53 releases, v0.41.0 to v0.61.0, newest first. The newest releases and every other month are on the changelog.
v0.61.02026-09-30
The memory handoff can be measured
—Ticking Measure whether it helps holds it back from one new session in four, at random, and the Memory screen compares the two groups: minutes and tool calls to a session's first edit. Off by default. On the owner's machine a session continuing yesterday's work reached its first edit no faster than one starting cold, with the handoff already on — this is how to find out whether it does anything.
v0.60.12026-09-30
Codex sessions on gpt-6-astra and gpt-6.1-sol are priced
—Codex sessions on gpt-6-astra and gpt-6.1-sol are priced (OpenAI list prices, read 2026-09-30). They showed as "Partial cost"; the turns already stored are priced on the next start. The gpt-6 rows also carry OpenAI's cache-write rate, which Codex does not report tokens for yet.
v0.60.02026-09-30
After a reboot or an OS update, the Now screen names the sessions…
—After a reboot or an OS update, the Now screen names the sessions that were still running when Caprock stopped, each with continue, until dismissed (FB-038). They used to be a few ended cards among a hundred, found by opening each.
—A /clear or a fork is one card, not several: the earlier parts of the conversation ride on the latest card as links (FB-039). Only recorded facts join sessions — the same process after /clear, or a fork Caprock started — never the same folder or branch.
—An ended card shows when the session was worked in — "worked 29 Sep 14:02–16:40" — not when it was closed. After a reboot every card read the reboot's time (FB-037), and the list is now ordered by the same time.
—A session from before a reboot is ended even when its old process number now belongs to another program. It stayed "live" instead, and its terminal tab was an empty screen.
—The terminal tab of a session Caprock started before it restarted says the terminal went with that run and offers continue, instead of an empty screen (FB-040).
v0.59.62026-09-30
Hook events are no longer lost when the database is busy
—The shim waits a second and hangs up, and an event still waiting for the write lock was abandoned with it — 41 on the owner's machine, each logged as "context canceled" and missing from the record. The write now finishes after the shim has gone.
—The log no longer grows without bound. A model missing from the pricing table was warned about on every event, re-reads included — 1.1M identical lines and a 4.6GB caprock.log. It is now said once per model, and the daemon's log is rotated at 64MB with one previous file kept; a log the daemon has stopped writing to is cut to its last 8MB.
—Restarting the daemon no longer re-reads every Codex transcript. On the owner's machine that was 1.1GB and 30k duplicate writes on every start, while the dashboard and hooks waited; only transcripts that changed are read now.
—CI checks the query plan of every aggregate the dashboard polls and fails when one would read full event rows instead of a covering index — the regression behind 0.59.2 and 0.59.3, which a small test database could never show by timing.
v0.59.52026-09-30
A session started from Caprock right after the daemon starts gets…
—A session started from Caprock right after the daemon starts gets your login shell's environment instead of the daemon's bare one. Reading it was cut off after 5 seconds, and at start — while every importer does its first pass — it could take longer; the failure left nothing to fall back on but launchd's PATH, so gcloud went missing again (FB-033). The limit is now 30 seconds, a failed first read retries on its own, and a session started meanwhile waits up to 15 seconds for it.
—caprock status says which environment sessions start with — your login shell's, still being read, or the daemon's with the reason — so a session that cannot find a tool a terminal can find is one line to diagnose.
v0.59.42026-09-30
Built with the modernc.org/sqlite 1.59 driver, and the dashboard with…
—Built with the modernc.org/sqlite 1.59 driver, and the dashboard with Vite 8.3.1. No change in behaviour; the summary and history timings were re-measured against the same database copy as 0.59.3 and match.
v0.59.32026-09-30
The Lifetime screen and the all-time figures on Now load in about a…
—The Lifetime screen and the all-time figures on Now load in about a second instead of nine. The indexes the spend aggregates read from stopped covering them when background usage was split out; migration 0026 rebuilds them (about ten seconds on a large database, once, on the first start). The History totals for today read off the time index like the summary's already do (4.8s to 0.02s on the owner's database).
—Rebuilding the Codex daily totals no longer fails on a busy first start with "database is locked"; it used to wait for the next restart.
v0.59.22026-09-30
Today's totals load in a fraction of a second again
—Two of the queries behind them walked every event in the database instead of today's, once the background-usage split added a column their indexes did not cover; on a 300k-event database the summary took 3.7s at rest and tens of seconds while a session was busy, and screens sat on "reading your figures…" or "loading…". Measured on a copy of the owner's database: today 3.7s → 0.5s, 7d 1.9s → 0.15s.
v0.59.12026-09-30
The dashboard no longer piles up requests while a session is busy
—Every live event made each panel ask again, whether or not its last request had answered, so on a large database today's totals fell 30-60 seconds behind and read "reading your figures…" while the daemon ran at over 100% CPU. A panel now waits for its running request and asks once more after it.
—make shots waits until every placeholder on a screen has cleared, and fails rather than capturing a half-loaded one.
v0.59.02026-09-30
Every session card says what the session was about
—Claude Code's, Codex's or OpenCode's own title for it (the name their resume pickers show), or else its first prompt that says something. Ended sessions can finally be told apart (FB-035). OpenCode sessions imported before this release get their titles too.
—Pricing for GPT-6 Sol and GPT-6 Luna, from OpenAI's published list prices (read 2026-09-29). Turns already stored without a price because their model was missing from the table are priced on the next start, at the rate in force when they ran; turns that were priced keep their figures.
—An ended session says whether it can be continued and, when it cannot, why: the transcript was deleted by Claude Code, the folder is gone, or the agent is not one Caprock resumes — with the command to resume it yourself where the agent has one (FB-036).
—Ended session cards say when the session ended and offer continue (or why not) right on the card. With ended sessions shown, a search box finds a session by its title, any prompt typed in it, its project or branch, and show more reaches past the first 200.
—A session started from Caprock can now reach what a terminal can: gcloud, bq, Homebrew tools and everything the shell profile exports. Sessions and verification commands used to inherit the daemon's environment, which under launchd autostart is a bare PATH=/usr/bin:/bin:/usr/sbin:/sbin; they now start with the user's login-shell environment, resolved in the background at start and refreshed in the background once it is more than a minute old (FB-033). Caprock runs your $SHELL -l -i to read that environment; the shell sees CAPROCK_RESOLVING_ENVIRONMENT=1, so a profile can skip interactive-only work. Windows is unchanged.
—A session started from Caprock no longer inherits Claude Code's session markers when the daemon itself was started from inside a Claude Code session.
—Pasting into the terminal sends the text once. Cmd+V (Ctrl+Shift+V off macOS) used to paste twice whenever the browser allowed a clipboard read, so short text appeared doubled (FB-034).
—A long paste collapses into Claude Code's "[Pasted text]" block again after switching tabs or reloading. A terminal that attached after the session's first 256 KB of output never learned that bracketed paste was on; the daemon now restores the terminal modes the session set before replaying scrollback.
—Sessions Caprock started can be continued after they end; the button used to disappear for them, including after they had been continued once.
—Continuing a session whose transcript is gone is refused with the reason, instead of opening a terminal that printed "No conversation found" and exited.
—Daily totals no longer count some Codex turns twice. Migrations 0022 and 0023 re-imported every Codex event without taking the first import out of the daily table; the Codex rows are rebuilt from events once on start (on the owner's machine, five days $3.68 and 7.9M tokens too high). A Codex turn repriced after its model became readable now moves the session and daily totals with it.
—Pasting an image with the keyboard uploads it and types its path; only drag-and-drop worked before, because xterm stopped the paste event from reaching our handler.
v0.58.32026-09-13
Refreshed the shipped UI's React runtime, React types, and Node…
—Refreshed the shipped UI's React runtime, React types, and Node types; the embedded production bundle is rebuilt and verified across all CI platforms.
v0.58.22026-09-13
Refreshed the SQLite and Go system dependencies and the pinned GitHub…
—Refreshed the SQLite and Go system dependencies and the pinned GitHub Actions used by CI, with the full cross-platform verification matrix still green.
—Dependabot updates are grouped by ecosystem and capped at three open PRs; npm major updates remain separate for deliberate review.
—The README pulse animation now advances every 60 ms instead of 250 ms, so a complete dark or light cycle takes about 5.5 seconds and reads as live motion.
v0.58.12026-09-12
Checked-in DeepSeek Harness transcript fixtures cover current and…
—Checked-in DeepSeek Harness transcript fixtures cover current and legacy JSONL formats without depending on a developer's ~/.dsh directory.
—Dependabot now watches GitHub Actions, Go modules, and ui/ npm dependencies.
—GitHub Actions are pinned to immutable commit SHAs with readable release comments, reducing supply-chain exposure while keeping updates automated.
—The incomplete-cost warning now says Partial cost, offers copy/report/why actions, and explains the lower-bound total only on demand. No manual price entry is required.
v0.58.02026-09-12
DeepSeek Harness is the fifth agent on the same screen
—Caprock now reads DSH's own session transcripts (~/.dsh/sessions/…, zstd JSONL) and shows its sessions, turns, tool calls and cost beside Claude Code, OpenCode, Gemini and Codex. The model DSH names — deepseek-v4-pro — was already in the pricing table, so no pricing change was needed. Observation-only, like Codex: Caprock reads the files and does not start, steer or stop DSH sessions.
v0.57.02026-09-11
The default port moved from 4173 to 22776
—4173 is Vite Preview's default, so a developer running vite preview beside Caprock hit a bind collision. A fresh install now listens on 22776 (unassigned by IANA, spells CAPRO on a phone keypad); an existing install that never wrote config.json keeps its old 4173 origin, because a port is part of a browser origin and moving it would strand bookmarks and LAN-pairing tokens.
—Codex's own approval reviewer was reported as user work. codex-auto-review is Codex's hidden "Automatic approval review model", and it writes a normal transcript that raised an unresolvable unpriced-cost warning (OpenAI publishes no price for the id). It is now classified as background machinery, kept out of every user-work total, and reported beside them as a quiet "background usage" line with no dollar value. A genuinely unknown model still gets the actionable "partial estimate" with a prefilled report issue.
v0.56.02026-09-09
The context tax, in the screens you already look at
—Every call re-sends the whole conversation as a cache read before it does anything, and Caprock stored every number needed to price that and showed none of them. It now does, in three places and no new ones: the session card's Context caption says what the next call costs at the current fill ($0.07/call), the lifetime Breakdown carries a Context tax row with its dollars and its share of what was priced, and a loop alert states what the repeated calls paid to re-read the conversation. On the author's own archive that row reads $11,117 of $13,910 — 80%.
—The recording stand outlived its terminal, forever. Eight of them were found still running, the oldest two weeks old, each holding a daemon and a temp directory open. sleep 3600 in a hold loop is a separate process, and bash will not run a trap until it returns — so a stand started detached never received the interrupt that would have stopped it. It waits on an interruptible sleep now. Affects make record only, not the daemon.
v0.55.02026-09-06
More words for a cache hit rate, and a shorter cost on the status line
—98% and 99% were both "outstanding"; there are now excellent (98–99) and very good (96–98) between it and good, which moves down to 90. And the status line's cost read $169.410 — a trailing zero on a line that reprints after every message. It now shows two decimals, matching the dashboard, and four only below a cent: one decimal was the obvious cut, but 92 of 250 real sessions round to zero at one place, and a session that cost money should never report nothing.
—The waiting banner said the same thing twice. It read Waiting on you · caprock · waiting for you — one sentence in two prepositions, the repeat in grey as though it were extra information. Now it says it once, as "Waiting for you", and the "what did it ask?" button beside it is the part that actually tells you something.
v0.54.52026-09-06
Codex figures already on your screen are corrected
—0.54.4 fixed how token reports are read and could not reach the turns already stored — those kept showing no tokens and no cost. They are re-read from the transcripts on the next start, which takes seconds and needs nothing from you.
v0.54.42026-09-06
Half of Codex's token reports were being thrown away
—About half the turns in a Codex transcript state how many tokens they used without breaking it down by kind, and Caprock read only the breakdown — so those turns were stored as having used nothing at all. On the owner's machine that hid 18.6M tokens and turned $23.71 of usage into $0.53.
v0.54.32026-09-06
A Codex transcript with one unexpected field was skipped entirely
—The session record was read all-or-nothing, so a single field of a type we did not expect discarded the session id — and with it every turn in that transcript. Each field is now read on its own: an unfamiliar shape costs that field, never the session. Codex is someone else's format and free to change between releases, so this is the difference between missing a detail and missing a day's work.
v0.54.22026-09-06
Codex sessions were missing more than half their turns
—The importer keyed each event on a field present in one transcript out of a hundred; in the rest it read as zero, so every turn of a session shared one key and all but the first were discarded as duplicates. One session kept 1 turn of its 55. On the owner's machine 98 turns were stored where the transcripts hold 218, with 18.5M tokens missing along with them.
v0.54.12026-09-06
Codex sessions now all carry a price
—v0.54.0 shipped with 83% of Codex tokens costing nothing, because it read the model from the obvious field — present in 4 transcripts out of 100. The other 96 record it elsewhere, as an explicit model id rather than a hint, so both are read now and every session that used tokens can be priced. A session that names no model anywhere still keeps its tokens and no cost, rather than an invented one.
—A test daemon read the machine's real Codex transcripts. The smoke test asserts exact turn and cost figures, and on a developer machine with Codex installed it was importing that person's actual sessions and counting them. It turns the import off, as it already did for OpenCode.
v0.54.02026-09-06
Codex sessions are on the dashboard
—A machine running Claude Code, OpenCode and OpenAI Codex had its spend split three ways, each tool knowing only its own share. Codex sessions, turns, tool calls and cost now sit on the same screens as the others, filterable by agent. Nothing to install: Codex writes one transcript per session and Caprock reads it, the way it already reads OpenCode's database.
—The share dialog's preview never changed when you switched period — and on a real machine it never appeared at all. Two faults in the same box. The draw released its image on teardown, so changing period released the one still on screen, and a released image leaves the previous picture showing. Under that, anything the drawing threw on killed the draw silently and left "drawing…" on screen for good, which looked exactly like a slow one. The card you saved was always right; the preview was a lie about it. A draw that cannot finish now says so.
—The "Lots of turns, few files" alert. It counted files touched through Edit/Write only, so everything changed through Bash — sed -i, a heredoc, a python one-liner — was invisible to it. An agent running with permissions bypassed is told to prefer exactly those, so the rule was blind to an ordinary way of working.
v0.53.22026-09-06
A status line that had silently printed nothing is repaired on start
—An older version wrote the registered command with quotes around the whole string rather than just the path ("…/caprock statusline"). The shell reads that as one filename, finds no such file, and prints nothing — so the status line was simply blank, with nothing anywhere saying why. Worse, Caprock did not recognise the entry as its own, so statusline install treated it as somebody else's and refused to touch it: the machine it was found on had been in that state for months. caprock up now recognises the spelling, repairs it, and says so. A statusLine that is not ours is still never touched, and neither is a working one.
v0.53.12026-09-06
Every coloured border in the dashboard was grey
—A global border reset sat outside Tailwind's cascade layers, and unlayered CSS beats layered CSS whatever its specificity — so it quietly overrode every border-accent, border-premium, border-ok and border-warn in the app. It went unnoticed because the elements using them mostly have a tinted background too: the "+ New session" button looked amber because of its fill, not its border. Buttons and status pills now show the colour they always asked for.
—The two session buttons line up with the strip beside them. Quick chat and + New session were shorter than the ALL TIME row they share, which read as ragged; they now match its height. Quick chat gains a soft amber border.
v0.53.02026-09-06
The status line carries the mark
—the amber ⛰ from the favicon, the glyph alone, so the line is identifiably ours instead of an anonymous row of figures. Two columns, and it is never dropped however narrow the terminal gets. This is on every status line, in both modes, and needs nothing turned on: if Caprock's status line is already set up, upgrading is enough.
—The session's own numbers in the Claude Code status line — N turns · N steps · cache −N% · in N · out N, from counters only the daemon has, added to the line Caprock already prints.
v0.52.32026-09-03
The loop banner sold a cap that could not act, beside a cost that was…
—The loop banner sold a cap that could not act, beside a cost that was not the loop's. Both halves of the row overstated what they knew. The spend cap pauses only sessions Caprock started (rule 7), and 122 of the 127 sessions that did real work on this machine were started by hand — so "a cap that stops this" was, nearly every time it appeared, an offer to stop something the cap cannot reach. It now says "a cap covers sessions Caprock starts" and offers "what a cap does" unless the session is one Caprock owns; unknown ownership counts as not owned, since a promise made on a guess is the thing being fixed. The button stays, because the feature is real and the sessions where it applies are exactly the ones someone would want it for.
v0.52.22026-09-03
Sessions cleared before 0.52.1 still said "compacting context"
—0.52.1 gave /clear and Escape their own event kinds, but only for events recorded after it shipped — the rows already in the database kept the compact's kind, and on this machine all 18 of them were the last event of their session, which is the phrase the session card shows. So 17 sessions went on claiming a compaction that never happened, and the /clear that started the whole investigation was still captioned "was compacting context" beside a $1,062 cost and a 95%-full context.
v0.52.12026-09-03
/clear showed one editor as two working sessions
—Claude Code does not end a session on /clear — it keeps the process and starts a fresh session id inside it. Both rows therefore shared a live pid, and the sweep that decides a session is over asks whether its process is still running, so the old row could never close. It sat on the dashboard as a second working session, frozen at the moment it was cleared, still presenting a six-day cost and a 95%-full context as current. The replacement is now keyed off the SessionStart that follows, which is the only one of the two hooks that names the session meant to survive.
—A cleared session was narrated as "compacting context". /clear was stored as context.compact, which is a different event: a compact summarizes the context and keeps its substance, /clear discards it. The dashboard reported a compaction that never happened, at the one moment the owner is deciding whether to wait. It has its own kind now, and the reasons that leave a session running without touching its context (Escape at the prompt) have theirs.
—"unknown model" appeared beside the model's own name. A session that had not answered yet has no prompt to measure, so the Context stat was empty — and captioned unknown model directly beside a Cost stat reading claude-opus-5. The two reasons a context can be missing call for opposite reactions (one clears itself in seconds, the other never will), so the server now says which applies, naming the model id when the pricing table cannot size it.
—Fable 5 could not be picked for a new session. The model list offered three ids and omitted the most capable one on the machine, though claude --help names fable alongside opus and sonnet. All four are now offered, ordered by price, each verified by a live answer from the real claude — claude-mythos-5 is in the pricing table and stays out, because the binary rejects it.
v0.52.02026-09-03
A switch for the session memory
—"Sometimes I need the context and sometimes I don't" — a fair objection from the first person to live in Caprock, who had been bitten by an assistant that remembered his job title and volunteered it while he was asking about something else.
—"Answers" is now "Memory". Two halves of one thing did not know about each other: a new session is handed what the last one left, and you can search everything the agents ever said. Same corpus — one was labelled "memory" in settings, the other "Answers" in the nav, which nobody looking for the first would open. The status row names the screen too.
—The Memory screen says how far back it goes — "since July 19", measured from the oldest passage still held rather than from the fortnight a handoff reaches. Those are two different dates, and showing the shorter one understated the corpus by a month.
v0.51.12026-09-03
The token column beside each model counted re-reading, not work
—"claude-opus-5 · 12.61B tokens" is arithmetically true and reads as how much went through that model. On a normal workload 98.6% of it is cache read — the same context fetched again, at a tenth of the price — against 14.6M of output. The column measured how many times one conversation was re-read.
v0.51.02026-09-03
A session opening a repository you have worked in before already…
—A session opening a repository you have worked in before already knows where you stopped. Open claude, and the agent has been handed the last substantial thing an agent said in that repository — no command, no prompt, nothing to press. Verified against a live session, which named the previous session's work while forbidden to read files or git.
v0.50.12026-09-03
The update instructions were set like captions
—The commands read fine; everything explaining them was 11px in the faintest grey — the weight this interface uses for a timestamp, something you are meant to slide past. It is the half of the dialog that says why updating takes three commands and why Caprock will not do it for you. The install-route tabs got the same lift; they are a choice, and they were dimmer than the platform tabs beside them.
v0.50.02026-09-03
Let a tablet in from the dashboard, without a terminal
—Reading Caprock from a phone shipped needing caprock up --lan, which requires a terminal on the machine — and the person who wants this is holding the tablet. It is a button on the status screen now, with an off switch beside the device list. The flag still works for a machine administered over SSH. Unchanged: it is off again when Caprock restarts, and paired devices are kept.
—Caprock uses a Tailscale address when there is one, and says which kind of address you have. A LAN address works when the other device is on the same wifi and never from mobile data — no setting changes that, because the packets do not arrive. The panel said nothing about it, so the honest conclusion from a tablet on mobile data was that the feature is broken.
v0.50.02026-09-03
Let a tablet in from the dashboard, without a terminal
—Reading Caprock from a phone shipped needing caprock up --lan, which requires a terminal on the machine — and the person who wants this is usually holding the tablet, not sitting at the machine. It is a button on the status screen now: press it, and the same panel shows the address and the code. The flag still works for a machine you administer over SSH.
v0.49.32026-09-03
"Burn now" says it is measuring instead of extrapolating
—0.49.1 made the rate divide by the time it actually covers, which is correct arithmetic and an alarming number: a daemon twenty seconds old showed $3,215/h, because twenty seconds of a busy minute extrapolates to that. Until the window has been lived through, the tile shows a dash and says it needs ten minutes of running — a figure nobody can act on is worse than no figure.
—A test that asserted a promise the code never made. One of the night's new tests demanded that a healthy terminal subscriber receive 300 of 400 chunks while another subscriber sat stalled. The buffer is deliberately lossy — a reader that falls behind legitimately misses some — so the test passed on a fast machine and failed in CI, which is what stopped v0.49.2 from publishing. It asserts what the code actually promises now: the writer finishes and a healthy reader keeps receiving.
v0.49.12026-09-03
"Files touched" counted whole session lifetimes
—The range filtered which sessions counted, not which files — so a fortnight-long session that touched 300 files put all 300 into "today" the moment it said anything today. Seven days read 417 against a true 144. The ranged answer was already on disk; the wrong table was being asked.
—The daily session count read zero on days with real spend. It counted a session on the day of its first turn ever, so a session begun yesterday added nothing to today. Four of eight days read 0 against 1, 2, 2 and 2 — beside cost and token figures in the same rows that were correct, which is worse than being wrong alone. After the migration, ten recent days all match a direct count over the events.
—A replaced transcript of the same size lost everything below the old offset. Truncation was detected by "the file got shorter", and Claude Code rewrites transcripts in place — on a resume, on a compact — where the replacement usually is not shorter. Those lines were never parsed, so nothing deduplicated them either: a session simply arrived missing its first turns.
—A half-written Gemini record was skipped permanently. The reader advanced past everything available whenever anything parsed, which is the opposite of what its own comment promised. Since the file is read while Gemini writes it, landing mid-record is the normal case.
—Two Gemini events in the same millisecond became one — and worse, every event whose timestamp was missing or in an unexpected format shared a single key, so only the first was ever stored.
—The burn rate divided a partial window by its full width. A daemon three minutes old has three minutes of history and was dividing by ten, so the rate read a third of the truth with nothing saying the window was still filling.
—Six tables can no longer push the page sideways on a narrow screen. On a tablet a wide table took the whole body with it.
—More tests where a failure would cost something: the three shapes a tool error arrives in (only one was exercised, and it decides the red badge), the point where rule 7 is enforced, all five pairing handlers, and the file that holds device tokens. Every important test was checked by reverting the behaviour it asserts.
v0.49.02026-09-03
Read Caprock from a tablet
—caprock up --lan opens a second listener on this machine's own private address; the status screen shows what to open on the other device and a six-digit code to type. No account, no tunnel, nothing on anybody's server — the other device is talking to your machine.
—Two figures ignored the agent filter. Choosing opencode on the Now screen narrowed most of the Today panel and left "burn now" showing every agent's money, in the same grid row as the cost it no longer matched. The unpriced warning had the same fault, and worse: it told a reader looking at one agent that tokens were missing from a total those tokens were never in.
—The share card's breakdown was always the last thirty days. Choosing a period changed the heading and the lit tile and nothing else that mattered, so a card headed "My week on Caprock" carried a month's models beneath it — five of them against the one that week ran, and $5,459 of Opus against $1,576. The four tiles still show every period on purpose; it is the breakdown that belongs to the period the card names.
—A range or filter change showed the previous answer under the new heading. Pressing 7d while 30d was up swapped the title at once and left every figure below it answering the old range until the response landed — around half a second on a large database, long enough to read a number and believe it. A refresh still keeps the figures on screen, because blanking them every few seconds would be its own defect.
—The calendar put days in the wrong weekday column. A day nobody worked has no row at all, so one missing Sunday slid every later day one column left and a Monday rendered under "S" — in a panel whose whole purpose is making "I do not work Sundays" visible at a glance.
—"Ended · 200" counted the page, not the sessions. The list is capped at 200, and the label counted the array it received while the lifetime strip above said otherwise. It reads "200 of 431" when they differ.
—The largest figure on the screen called itself a cost. On a flat plan it is not one: it is what the work would have cost through the API, and the qualifier saying so sat underneath in grey at a third the size. A reader on a subscription took the whole product for an accusation of waste. The figure is unchanged and now headed "Worth at API list"; anyone billed per token still sees "Cost", because for them it is the true word.
—Every screenshot retaken, and three faults in how they are captured fixed: two first-run banners were in every frame, "set plan" was lit as an unfinished action because the capture runs against a throwaway data directory, and the feature crops had silently been failing to find their panels since the database grew.
—The docs say what the product does now. An audit found the release check described as the only outbound call (it is one of three, all opt-in), the premium page selling a shipped feature as unbuilt, a release runbook asking you to verify a hook count that can no longer appear, and Gemini missing from every user-facing list of agents.
v0.48.02026-09-02
Every table says what its columns are
—Three figures in a row with nothing over them is a puzzle: $7,587.14 · 12.01B · 58% reads as three unrelated numbers until you work out which is which, and the "by cost" in the corner describes the sort order rather than the columns. Every breakdown on Now, Cost and Lifetime now names them, on one line with the table's own title so it costs no vertical space.
—The all-time breakdown moved up, under the live pulse. It was last, below a tall activity feed, so the most legible thing on the screen was the thing you had to scroll to find.
—"Nothing measured yet" appeared while the figures were still loading. The first screen could not tell "no answer yet" from "the answer is nothing", so for the second it took the first response to arrive it announced, with confidence, that a machine with thousands of turns on it had never run anything. It says "reading your figures…" until it knows. Measured on a 641 MB database: nothing on this screen takes longer than 0.7s, so the delay was never the problem — the wrong answer during it was.
—The answers screen showed raw Markdown. Claude writes bold text, lists and tables; the screen kept the line breaks and showed everything else as the markup it is made of — bold with its asterisks, a table collapsed into | | |. The same renderer the release notes got now covers both, and it learned tables and quotes on the way.
—A plan you had already chosen looked unchosen. Once set, the picker's button went grey and lost its border — which is exactly what the buttons beside it look like when they are not set — so every session started with the impression that the plan had been forgotten again. It keeps its border and carries a check.
—A Gemini error sent people somewhere they no longer need to go. "Set GEMINI_API_KEY in the daemon's environment and restart it" — but the key has a field on the Cost screen and takes effect at once. A wrong instruction is worse than a long one.
—Errors carried a Go package name into the dialog box. agents: "/x" is not an absolute path was written for whoever was debugging it.
—The tool table says how much each tool hands back. Calls alone told the wrong story. Bash is 46,343 calls returning 13.4 MB; Read is 11,767 calls returning 58.7 MB — four times fewer calls, four times more context. The loud tool is the cheap one, and nothing in the product said so.
—Share a week, a month, or today — not only all time. The picker was removed once, on the reasoning that a card showing every period at once made choosing redundant. It does not: somebody sharing a working week does not want their lifetime total to be the headline, and a card that answers four questions answers none of them loudly. The choice now decides what the heading claims and which figure is lit; every other period stays on the card, because a week means nothing without knowing whether it was a normal one. It opens on the week.
—You can see the card before you send it. The buttons said where the picture would go and nothing said what was in it: you pressed Save, opened your downloads, and only then found out. The card is drawn in the sheet now, at the size it will be seen, and it redraws as you change the period — which is what makes the choice worth having.
—The Share button can be found. It was an outlined button in a row of 11px text, next to a solid premium button — an outline cannot win an argument with a fill. It is filled, a size up, and carries an icon. It earns that: sharing is the only thing on this dashboard that leaves the machine.
—Copy across the product says less. Several tooltips explained the attribution algorithm where the reader had asked what a number meant, and several paragraphs defended design decisions nobody had questioned. The longest of them — the per-directory rule — was four sentences of specification on hover; it is now one. "Everything here is measured — no invented numbers" is gone from the Cost screen: readers assume the figures are real until told otherwise, so the sentence created the doubt it answered.
—Messages about a missing claude say what to do. "Spawning is unavailable" named an internal verb and diagnosed without a fix; it now says the binary is not on PATH, that installing Claude Code fixes it, and that watching works regardless.
v0.47.02026-09-02
Pick up a session Caprock did not start
—It never types into a process it did not launch — two writers on one terminal interleave characters and ruin both — so a session you started yourself was readable here and nothing else.
v0.46.02026-09-02
Send this week's report now
—There was no way to find out whether a bot token and chat id were right except to wait for Monday — and when nothing arrived, that looked exactly like a week where nothing moved. A button beside the fields sends one immediately.
—What's new showed raw Markdown. ### Fixed as a literal heading, asterisks around every bold phrase, and paragraphs broken at the eighty columns a changelog is wrapped to for the sake of diffs. It was rendered as preformatted text on the reasoning that a local-first tool should not render remote markup — right instinct, wrong conclusion, in a dialog whose only job is to be read.
—The dialog linked to GitHub. Someone reading what they just upgraded into wants the changelog written for readers, not a tag page with build artefacts on it. It goes to caprock.dev/changelog now.
v0.45.22026-09-02
Why a weekly report never arrived is remembered now
—The reason a send failed was held in memory, so restarting the daemon erased it — and restarting is exactly what somebody does when a feature seems broken, which made investigating the problem the thing that destroyed the evidence.
v0.45.12026-09-02
The premium lock covered the sentence explaining the feature
—The caption and the button were centred on the panel, so they landed exactly across the middle of the text — "Off. Nothing is paused, whatever the day costs" was cut mid-word by the thing selling it. Two earlier attempts tuned the preview's opacity, which could not help: the problem was never how transparent the overlay was, it was where it sat. It is a row underneath now, reading in the order the eye already moves — here is the feature, here is what it costs — and covering nothing.
—A gemini filter chip appeared on a machine with no Gemini sessions. Having the binary on PATH is what decides whether the New Session dialog offers Gemini; it says nothing about whether any Gemini session exists. A first run showed all · claude · gemini above an empty screen with nothing to sort by. The chip now appears when there is something to filter.
v0.45.02026-09-02
A session stays open until its process exits
—It used to close after a stretch of silence, and every number chosen for "a stretch" was wrong for somebody: twelve hours left the day's work marked live at midnight, and one hour closed a session while its owner was at lunch. Both were guesses about a person's day standing in for a fact about a process — and the fact is available. The shim now reports the pid of the Claude Code that ran it, and Caprock already knew the pid of every session it starts itself, so the sweep asks whether that process is alive instead of watching a clock.
—The Now screen shows now. A session that has not made a sound in two days no longer appears there, whatever its status column says — status alone cannot tell a quiet session from somebody's history, because an observed agent's sessions arrive weeks old by definition. Two days rather than one, so a session left on Friday evening is still on screen on Monday morning. Anything actually working shows regardless of age, which is what the screen is for. Older sessions are on Lifetime and under "show ended".
v0.44.52026-09-02
A session you left for lunch was closed while you were out
—0.44.3 cut the staleness threshold from twelve hours to one, on the reasoning that an hour outlasts lunch. An hour is lunch: the first person to leave a terminal and come back to it found the session gone from the dashboard. On the machine this was checked against, 44 sessions had paused for over an hour and then carried on working, and 86 of those pauses were between one and three hours — a meeting, a walk, an afternoon on something else.
v0.44.42026-09-02
Caprock now sees what a Gemini session does
—It could start one and then observe nothing: a live session sat in the list with zero turns, zero tokens and no cost. Gemini has no hooks and writes no transcript, but it does write OpenTelemetry, and it will write it to a file Caprock names — so a spawned session now reports its turns, its tokens and its cost like any other.
—Two of the three Gemini models still could not be used. The list was checked against the installed CLI and the pricing table, and both said these were fine. A real key says otherwise: Google has closed gemini-2.5-flash-lite to new keys, and gemini-3.1-pro-preview is barred on a free key by quota (generate_content_free_tier_input_token_count, limit: 0) — so picking either opened a terminal and then failed.
v0.44.32026-09-02
A session you were still working in could close itself
—0.44.0 started listening to Claude Code's SessionEnd hook so a session ends the moment it ends rather than whenever the staleness sweep next runs. But SessionEnd does not only mean "the user left": it also fires on /clear, on Escape at the prompt, and with an unhelpful other. All of them were being treated as the end. One session on the machine this was found on had been running for six days when a /clear retired it in the dashboard.
v0.44.22026-09-02
The terminal flickered on every keystroke
—A ResizeObserver called fit() straight from its callback, and fit() writes to the element the observer is watching — so a resize could cause a resize. Idle, that settles after one frame and nobody notices. Under a TUI that repaints its whole box on every key, which Gemini CLI does, it was a visible flicker on every key. Fits are now coalesced into one per frame and skipped entirely when the geometry has not moved. It affected any terminal that repaints often, not only Gemini's.
—Six columns of zeros above a session Caprock cannot measure. A Gemini session showed $0.00, 0 tokens, 0 turns, 0% cache, with the reason — "no hooks · no transcript" — in small grey text underneath. Zeros are how that bar says "nothing has happened yet", so it read as broken rather than as out of scope. It now says the one true thing instead: Caprock started this session but does not measure it, and the terminal below is live.
v0.44.12026-09-02
Starting a session with Gemini actually works now
—0.44.0 shipped the feature built against a fake binary, on four assumptions about the Gemini CLI that were written from memory rather than from gemini --help. Installing the real thing disproved all four in ten minutes.
—A Gemini session is no longer described as a Claude one. The agent was added to the launcher and to nothing else: its sessions arrived in the list wearing no badge, with no filter chip to find them by, and the empty state called them Claude Code — because three separate places asked agent === 'opencode' and treated everything else as Claude. There is now a gemini chip beside claude and opencode, shown only on a machine that has it, and one helper decides the badge and the name so a fourth agent cannot reintroduce this.
v0.44.02026-09-02
Start a session with Gemini CLI
—Install it (npm install -g @google/gemini-cli), and it appears next to Claude Code in the New Session dialog — same terminal, same directory picker, same row in the sessions list, same cost stream. Caprock passes the key you already entered into the child process, so it works without exporting anything in the shell you happen to be in.
—A debounce timer outlived the screen that set it. Closing a screen inside the 400 ms live-tick window left a timer that woke up and set state on a component that no longer existed. It was visible mainly as a failed release — the timer firing after the test runner had torn the DOM down, with all 483 tests passing — which is how it got fixed twice at the wrong end before the cancel-on-unmount that it actually needed.
v0.43.02026-09-02
Claude Fable 5.1 and Mythos 5.1 in the pricing table, contributed by…
—Claude Fable 5.1 and Mythos 5.1 in the pricing table, contributed by [@SergeAx](https://github.com/SergeAx) — the first outside contribution to this repository. Cache reads on those two models cost 0.025× the input price rather than the 0.1× every other Claude row follows, which makes $0.25 look like a typo of Fable 5's $1.00; the PR pins it with a test, so a later tidy cannot quietly charge four times the real rate.
—fetched_at in the pricing table names the most recent read, and each note carries the date for the rows it describes. One date on a table assembled from three vendors over three weeks said 18 August while holding prices read on 1 and 2 September — which tells a reader checking a figure nothing about that figure.
—The Gemini key goes in the dashboard now. It was read only from GEMINI_API_KEY, on the reasoning that a key Caprock never holds is one it can never leak — but a daemon started at login inherits nothing from a shell profile, so switching on a paid feature meant hand-editing a launchd plist. A key nobody can enter protects nothing; the feature just goes unused. Paste it into the panel instead. The environment variable still works and still wins where both exist, and the stored key is never returned by any endpoint — the same write-only rule the Telegram token follows.
v0.42.12026-09-01
The weekly report, to your own Telegram bot
—What moved this week against your usual, sent Monday morning — or the next day you open the lid, because a timer set for Monday 09:00 fires for nobody whose laptop is shut at weekends.
—A setting no longer resets when you save a different one. "Set plan keeps resetting" was not a save that failed: the dashboard sent the whole settings object every time, assembled from a cache in the page, so any control could write back a field it had read minutes earlier. Two tabs were enough — change the plan in one, turn on update checks in the other, and the second restated the plan from its own stale copy. Only what changed is sent now.
—Live activity showed a fortnight of history as what just happened. Two causes at once: the feed asked for a session's first sixty events rather than its last, and the store ordered "most recent" by insertion rather than by time. Those agree for a session captured live and diverge for one whose transcript was re-read, because a backfill inserts old events with new ids. On a long session every row was equally stale, which is also why it read as impossible to scroll.
—The new-session dialog cut off the permission label — "Accept edits · asks before com…" hid the consequence, which is the one thing that label exists to state. Wider dialog, shorter labels.
—A stray reconnect no longer crashes a torn-down page. The live socket queues a retry when it cannot connect, and that timer could outlive the document that scheduled it — harmless in a browser, where the page is going away anyway, and enough to fail a release build whose every test had passed. It now gives up quietly instead.
—The premium chip was two controls disguised as one. The label went straight to the checkout while a chevron beside it opened the explanation, in one border with no seam between them — so clicking the word "premium" to find out what premium is took you to a card form. One control now: it opens the dialog, and once you have bought it names your plan and when it ends.
v0.41.02026-09-01
Ask Gemini, on your own key
—The second paid feature: a second model inside Caprock, billed to you by Google at their prices. Caprock never stores the key — it reads GEMINI_API_KEY from the daemon's environment at the moment of the call, and it is never written to disk, never accepted by the settings endpoint and never returned by one. A key we do not hold is a key we cannot leak; the cost is that you set a variable and restart, which the panel says plainly.
Windows 10 22H2 or later, x64. Windows may ask once whether to run it: click More info, then Run anyway.
Paste it into Claude Code, Codex or any coding agent.
See what it says ▾
Install the Caprock desktop app for me. It is open source (Apache-2.0), a local
app for my coding agents: https://github.com/dspv/caprock
What the install does and how to remove it:
https://github.com/dspv/caprock/blob/master/docs/install-agent.md
Run the install script for my system: https://caprock.dev/mac.sh, /win.ps1 or
/linux.sh. Read it before you run it if you like. Leave my Claude Code settings
alone; the app asks me itself. Tell me when it's open.
Or run it yourself:
$curl -fsSL https://caprock.dev/mac.sh | shCopy
Paste in Terminal: it installs Caprock and opens it.Or with Homebrew: brew install --cask dspv/tap/caprock-app
Or run it yourself:
>irm https://caprock.dev/win.ps1 | iexCopy
Paste in PowerShell: it installs Caprock for you and opens it.