codeburn/docs
Resham Joshi 5d12c6e86c
feat(guard): warn and stop identical-call loops and failing-command retries (#1857)
* feat(guard): warn and stop identical-call loops and failing-command retries

The PreToolUse hook now watches two more signals next to the dollar caps.
The same tool called with the same input (hashed as JSON with sorted keys)
4 times in a row gets a one-time warning and the 8th call is denied. A shell
command (Bash or PowerShell) that already failed 3 times in the session gets
a one-time warning and is denied once it has failed 5 times; successful runs
do not reset the count. Both pairs are guard.json settings (loopWarnAfter,
loopStopAfter, retryWarnAfter, retryStopAfter), validated like the dollar
caps, null turns one off, and guard status prints them.

The incoming call is read from the hook payload and recorded by tool_use_id,
since the transcript can lag. The session cache (version 4) keeps a rolling
window of the last 16 calls, pending shell calls and failure counts per
command, folded from the same incremental tail read as the cost. When a loop
or retry warning fired, the Stop note adds one line with the count and what
the calls after the first warning cost.

On a generated 20 MB transcript the warm hook takes about 13 ms (median of 5);
the first cold read of the whole file takes about 480 ms, as before.

* fix(guard): let allow lift loop blocks and tell the agent at the warning

codeburn guard allow now lifts the loop and retry stops as well as the hard
cap, for that session only, so polling a CI job on purpose has the same way
out. The deny reasons name it.

Loop and retry warnings now also reach the agent through PreToolUse
additionalContext, with the same sentence plus "Change approach or ask the
user." It is sent without a permissionDecision, since "allow" would also
skip the user's permission prompt. The docs now say the loop count tops out
at 17 (16 earlier calls plus the incoming one).
2026-11-04 19:29:23 +01:00
..
design feat(dock): one Claude ring per config directory, with its own sessions and today (#1570) 2026-09-28 19:06:53 +02:00
providers chore: drop the no-zstd fallbacks the 22.15 floor made unreachable (#1855) 2026-11-04 19:29:18 +01:00
release-acceptance test(perf): add Phase 0 wait-path harness and pin main baselines 2026-08-29 23:45:46 +05:30
sync feat(sync): additive CB-3 usage-span fields and coverage watermark (Teams boundary spec section 7) (#1135, #1140) (#1147) 2026-08-25 11:32:45 -07:00
architecture.md feat(providers): add Command Code (#1658) 2026-10-07 00:18:50 +02:00
by-branch.md Add spend-by-branch report (spend --format branch-json) 2026-09-12 13:09:00 -07:00
cli.md Track peak vs off-peak usage for first-party DeepSeek and Z.ai routes (#1561) 2026-09-28 18:48:38 +02:00
compare-cohorts.md Wire cohort samples to the real drill-through, scope and serve the new flags 2026-09-12 14:02:01 -07:00
compare-periods.md Compare periods: lead sentence, three tiles, day bars, top movers (#1446) 2026-09-15 23:12:12 +02:00
configuration.md README: tell the story first, keep the manual in docs (#1527) 2026-09-22 16:28:27 +02:00
drill-through.md Fix drill-through provider, day and model attribution 2026-09-12 12:22:19 -07:00
guard.md feat(guard): warn and stop identical-call loops and failing-command retries (#1857) 2026-11-04 19:29:23 +01:00
how-it-works.md chore: drop the no-zstd fallbacks the 22.15 floor made unreachable (#1855) 2026-11-04 19:29:18 +01:00
mcp.md README: tell the story first, keep the manual in docs (#1527) 2026-09-22 16:28:27 +02:00
menubar.md feat(dock): one Claude ring per config directory, with its own sessions and today (#1570) 2026-09-28 19:06:53 +02:00
optimize.md README: tell the story first, keep the manual in docs (#1527) 2026-09-22 16:28:27 +02:00
plans-and-quota.md feat: Google AI plan presets and the Antigravity tier label (#1683) 2026-10-07 23:34:49 +02:00
README.md feat: CodeBurn for VS Code (#1670) 2026-10-07 16:46:23 +02:00
vscode.md chore: raise the Node floor to 22.15 and make zstd unconditional (#1846) 2026-11-04 17:37:28 +01:00
web.md feat(web): --allowed-host for an authenticating reverse proxy (#1789) 2026-11-02 17:17:06 +01:00
yield.md README: tell the story first, keep the manual in docs (#1527) 2026-09-22 16:28:27 +02:00

Docs index

One line per file in this directory, recursively.

docs/

  • architecture.md — A map of the codebase.
  • by-branch.md — The Spend view in CodeBurn Desktop has a By branch panel: pick a project and see where its AI spend went, per git branch, with the recorded worktrees and the individual sessions behind each branch.
  • cli.md — Every CodeBurn command and keyboard shortcut, plus the report-focused flags for filtering, JSON output, and diagnosing detection.
  • compare-cohorts.md — The classic Compare view answers "how do these two models differ across my whole history?" by aggregating everything each model ever did.
  • compare-periods.md — Compare two date ranges side by side and see exactly what drove the change in usage and cost.
  • configuration.md — Currency, model aliases, local/custom-priced/proxy models, and environment variables.
  • drill-through.md — CodeBurn Desktop can open any aggregate as a session list that explains exactly what composes it, without losing your place.
  • guard.md — Opt-in Claude Code hooks that watch session cost while you work: soft cap, hard cap, checkpoint nudge, session openers.
  • how-it-works.md — Pricing, task categories, breakdowns, one-shot rate, per-tool data locations and parsing, and how to read the dashboard.
  • mcp.md — codeburn mcp runs a local MCP server over stdio so an AI agent can ask about your usage and savings mid-conversation.
  • menubar.md — The macOS menu bar app, Windows tray app (with WSL discovery), and Linux GNOME Shell extension.
  • optimize.md — codeburn optimize scans your Claude Code sessions and your ~/.claude/ setup, reports what is costing tokens without earning them, and grades the setup A to F.
  • plans-and-quota.md — Subscription plan tracking and live provider quota via codeburn quota.
  • vscode.md — The VS Code extension: how it reuses the CLI client and the desktop renderer, which Node it runs on, workspace scope, and how to build and test it.
  • web.md — The local browser dashboard and combining usage across your paired devices.
  • yield.md — codeburn yield correlates AI sessions with git commits to show what actually shipped.

docs/design/

  • design/capacity-dock.md — Status: implementation specification (Epic: Quota Intelligence #725).
  • design/codeburn-desktop-plan.md — Goal: a standalone Electron desktop app that renders the approved v6 "indigo instrument" wireframes, fed entirely by spawning the codeburn CLI for JSON.
  • design/codeburn-desktop.md — Build CodeBurn Desktop, a standalone, resizable native desktop application (Electron) that renders the approved v6 "indigo instrument" wireframes.
  • design/codeburn-mcp-plan.md — Goal: add a codeburn mcp stdio MCP server exposing CodeBurn's usage/cost data to AI agents via two tools (get_usage, get_savings).
  • design/codeburn-mcp.md — CodeBurn already aggregates rich AI-coding usage/cost data (by task, model, project, provider; retry tax; routing waste; optimize findings; 365-day history).
  • design/desktop-data-lifecycle.md — Status: implementation contract for the quality/performance epic.
  • design/perf-cache-fix.md — codeburn status --format menubar-json was measured taking 25-90+ seconds per call, with no speedup on a repeat call against an unchanged, freshly-warmed cache.

docs/providers/

docs/release-acceptance/

docs/sync/

  • sync/README.md — Push your AI usage telemetry to a shared backend so teams can track adoption, budgets, and ROI across developers.
  • sync/DEVELOPER.md — Architecture, protocol, server contract, and testing for codeburn sync.