Skip to content

Improve agent token UX and blocked dependency handling #37

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
unbraind merged 2 commits into from
May 21, 2026
Merged

Improve agent token UX and blocked dependency handling #37

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7c4acd0
Improve agent token UX and blocked dependency handling
unbraind May 21, 2026
c059e65
Address health summary review feedback
unbraind May 21, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions .agents/pm/features/pm-rnpb.toon
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,15 +6,15 @@ status: in_progress
priority: 0
tags[10]: agent-ux,audit,calendar,ci,dogfood,pm-cli,sdk,search,security,telemetry
created_at: "2026-05-03T13:20:01.343Z"
updated_at: "2026-05-21T04:42:33.995Z"
updated_at: "2026-05-21T17:53:33.794Z"
deadline: "2026-06-01T00:00:00.000Z"
author: codex-audit
estimated_minutes: 300
acceptance_criteria: Audit findings are actionable and deduplicated; changed files and verification commands are linked; temporary dogfood project validates core workflows; any discovered bugs with safe fixes are remediated and tested.
parent: pm-jrjt
dependencies[1]{id,kind,created_at,author,source_kind}:
pm-jrjt,parent,"2026-05-03T13:20:01.343Z",codex-audit,null
comments[159]{created_at,author,text}:
comments[162]{created_at,author,text}:
"2026-05-03T13:20:01.343Z",codex-audit,Parent lineage established under pm-jrjt after duplicate checks found only closed prior audit items.
"2026-05-09T11:12:26.232Z",codex-agent,"Duplicate check before this pass: pm context --limit 10 --depth standard, pm search dogfood audit extensible sdk calendar semantic search telemetry sentry --limit 10, pm list-open --limit 20, and pm list-in-progress --limit 20. Reusing open feature pm-rnpb; prior same-day task pm-m35h is closed, so no duplicate item created."
"2026-05-09T11:46:59.430Z",codex-agent,"Evidence 2026-05-09: temp dogfood project exercised init/create/claim/update/files/docs/tests/calendar/context/aggregate/dedupe/search/reindex/extension scaffold/install/doctor/contracts/manage. Fixed search gap discovered there: reminder/event metadata is now in keyword and semantic corpus, and short phrase scoring ranks exact calendar event titles first. Verification passed: focused search/reindex/cache tests, release-readiness/search focused tests, linked pm test, full 100% coverage, static quality gate, CI workflow contract tests, pm health, pm validate, npm pack dry-run, secret scan, live telemetry/Sentry checks, and real hybrid reindex of 679 items."
Expand Down Expand Up @@ -174,6 +174,9 @@ comments[159]{created_at,author,text}:
"2026-05-21T04:17:16.756Z",codex-full-manual-audit,"Sub-agent/manual audit findings mapped without duplicate items: reuse pm-5k2w for config/calendar/telemetry/dependency/gc polish, pm-59gj for multi-package install/catalog UX, pm-rnpb for semantic provider staleness and auto-index provider parity, pm-orrl for blocked-by semantics, pm-nbht for health brevity, and pm-gt82/pm-mbdu for startup and large-file follow-ups."
"2026-05-21T04:40:06.326Z",codex-full-manual-audit,"Evidence: latest sub-agent/manual audit and fixes completed under pm-5k2w. Dogfood passed normal 84 workflows and semantic 86 workflows using qwen3-embedding:0.6b; release gate passed with coverage 96 files / 1399 tests at 100 percent and required Sentry/telemetry gate. Residual performance/coverage breadth/semantic provider-staleness findings remain mapped to existing pm-rnpb/pm-gt82/pm-mbdu/pm-59gj follow-ups, not duplicated."
"2026-05-21T04:42:32.110Z",codex-full-manual-audit,"Audit closeout evidence: broad manual dogfood and sub-agent review completed for this slice. Local release gate passed with telemetry required, package-first dogfood passed including semantic qwen Ollama mode, Sentry/telemetry gate passed against current baseline, GitHub issue/PR/vulnerability scan found no open actionable GitHub items. Residuals remain tracked under existing lineage: broader read-path performance, semantic provider staleness/auto-index parity, CI duration reduction beyond current selected coverage gate, and telemetry queue backlog investigation."
"2026-05-21T17:23:31.764Z",codex-gpt5,"Execution start 2026-05-21: bootstrap passed (npm install -g ., pm --version 2026.5.18, node v25.9.0, pnpm 10.33.0, pnpm build). Duplicate checks used pm context/search/list-open/list-in-progress; reusing active parent pm-rnpb and existing open children pm-6vfg/pm-nbht/pm-orrl, no duplicate items created. Dogfood finding: stale habit failed; runtime-discovered replacement is , and clearly reports allowed values brief|standard|deep."
"2026-05-21T17:42:51.746Z",codex-gpt5,"Final evidence 2026-05-21: completed child items pm-6vfg, pm-nbht, and pm-orrl without creating duplicates; closed and released all three with resolution/expected/actual metadata. Implemented MCP comments paging, health --summary, and create --blocked-by dependency semantics. Privacy pass used pm history-redact for pm-ny6y/pm-r9ei/pm-vhdc and verified each chain plus tracked secret scan. Verification passed: combined focused suite 8 files / 310 tests; manual temp smoke for blocked-by and health summary; package-first dogfood 84 workflows; semantic dogfood 86 workflows with qwen3-embedding:0.6b; full release gate 96 files / 1406 tests with 100 percent coverage and required Sentry/telemetry gate (Sentry critical=0 high=1 total=1 under threshold, telemetry finish_error_rate_pct=3.04, missing error-code rows=0). Resolution validation is clean; known legacy history drift remains under pm-xk39."
"2026-05-21T17:53:33.794Z",codex-gpt5,"Correction 2026-05-21: an earlier evidence comment had shell command substitution strip command names. Intended dogfood note: stale habit \"pm show --full\" failed; runtime-discovered replacement is \"pm get <id> --depth deep\"; invalid \"pm get <id> --depth full\" reports allowed values brief|standard|deep."
notes[2]{created_at,author,text}:
"2026-05-17T07:36:59.474Z",codex-agent,"Design note 2026-05-17: unfiltered pm contracts default now intentionally omits three heavy sections (schema oneOf union, command_flags, commander_aliases) and emits explicit omitted_reason markers; --full restores complete schema and alias/flag surfaces for tooling that needs maximal detail."
"2026-05-17T08:33:25.981Z",codex-agent,"Design note 2026-05-17: default brief projection was intentionally scoped to bare 'pm list' instead of all list-* commands to preserve existing automation contracts while still cutting agent token usage on the highest-frequency exploratory list call."
Expand Down
12 changes: 12 additions & 0 deletions .agents/pm/history/pm-6vfg.jsonl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1 +1,13 @@
{"ts":"2026-05-20T20:45:19.948Z","author":"claude-code-agent","op":"create","patch":[{"op":"add","path":"/metadata/id","value":"pm-6vfg"},{"op":"add","path":"/metadata/title","value":"MCP pm_comments returns full comment history (no default limit) — token bloat on long-lived items"},{"op":"add","path":"/metadata/description","value":"Calling MCP pm_comments on a long-lived item (e.g. pm-rnpb has 137 comments) returns the entire comment array (~92KB), exceeding the agent tool token cap and spilling to disk. Unlike pm_list/pm_search/activity (where a compact projection keeps all rows but trims fields), limiting comments DROPS older comments (lossy), so a naive default limit risks hiding history from agents that need it. The CLI `comments --limit N` exists. Decision needed on the right agent-default: (a) default to the most recent N with an explicit total count + hint to page older, (b) a compact projection that truncates each comment body, or (c) require explicit options.limit and document it in the tool description. Separately, a minor parsing quirk: passing options.add='text=...' with commas in the value stored the literal 'text=' prefix instead of stripping it."},{"op":"add","path":"/metadata/type","value":"Issue"},{"op":"add","path":"/metadata/status","value":"open"},{"op":"add","path":"/metadata/priority","value":2},{"op":"add","path":"/metadata/tags","value":["agent-ux","comments","dogfood","mcp","token-efficiency"]},{"op":"add","path":"/metadata/created_at","value":"2026-05-20T20:45:19.948Z"},{"op":"add","path":"/metadata/updated_at","value":"2026-05-20T20:45:19.948Z"},{"op":"add","path":"/metadata/author","value":"claude-code-agent"},{"op":"add","path":"/metadata/acceptance_criteria","value":"MCP pm_comments has a documented, non-lossy-by-surprise default that keeps agent output within token limits while preserving access to full history; tool description documents limit/paging; the add='text=...' prefix-stripping quirk is fixed or documented; regression test added."},{"op":"add","path":"/metadata/parent","value":"pm-rnpb"}],"before_hash":"3cc22dff72be7b14824654a7a64ea62b04799939b2fee54c1b5f52ca60bf6df0","after_hash":"3256e2d869ecc9a05b6a8df156e93ad9e00741095b9599bcf2ddd8c834b442cc","message":""}
{"ts":"2026-05-21T17:23:29.871Z","author":"codex-gpt5","op":"claim","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:23:29.871Z"},{"op":"add","path":"/metadata/assignee","value":"codex-gpt5"}],"before_hash":"3256e2d869ecc9a05b6a8df156e93ad9e00741095b9599bcf2ddd8c834b442cc","after_hash":"5ab977f80499f814c2298780b165e1f4de26ea2b0d528b76e17a01263d109f51"}
{"ts":"2026-05-21T17:23:33.468Z","author":"codex-gpt5","op":"update","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:23:33.468Z"},{"op":"replace","path":"/metadata/status","value":"in_progress"}],"before_hash":"5ab977f80499f814c2298780b165e1f4de26ea2b0d528b76e17a01263d109f51","after_hash":"ff4b83cb181a243a88561693039ca907b04fd974894c586a7fb6a8b5e59f3096","message":"Start MCP comments token-efficiency implementation"}
{"ts":"2026-05-21T17:23:35.608Z","author":"codex-gpt5","op":"comment_add","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:23:35.608Z"},{"op":"add","path":"/metadata/comments","value":[{"created_at":"2026-05-21T17:23:35.608Z","author":"codex-gpt5","text":"Execution start 2026-05-21: duplicate check reused parent pm-rnpb and existing child pm-6vfg after pm context/search/list-open/list-in-progress. Working on MCP comments default token-efficiency and add text= parsing quirk; sub-agents are reviewing related health/dependency/privacy surfaces."}]}],"before_hash":"ff4b83cb181a243a88561693039ca907b04fd974894c586a7fb6a8b5e59f3096","after_hash":"863f82451bd3cf599fb6fa9376288478b04d354797e6c9b58ce7ca1396add3cf"}
{"ts":"2026-05-21T17:30:24.024Z","author":"codex-gpt5","op":"files_add","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:30:24.024Z"},{"op":"add","path":"/metadata/files","value":[{"path":"src/mcp/server.ts","scope":"project","note":"MCP comments default paging and tool description"}]}],"before_hash":"863f82451bd3cf599fb6fa9376288478b04d354797e6c9b58ce7ca1396add3cf","after_hash":"7a77b61c49e7c50472650208d03926b20b702cb5b84ddf41280e3f254c986c83"}
{"ts":"2026-05-21T17:30:26.253Z","author":"codex-gpt5","op":"files_add","patch":[{"op":"replace","path":"/metadata/files/0/note","value":"comment paging metadata and text= parsing"},{"op":"replace","path":"/metadata/files/0/path","value":"src/cli/commands/comments.ts"},{"op":"add","path":"/metadata/files/1","value":{"path":"src/mcp/server.ts","scope":"project","note":"MCP comments default paging and tool description"}},{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:30:26.253Z"}],"before_hash":"7a77b61c49e7c50472650208d03926b20b702cb5b84ddf41280e3f254c986c83","after_hash":"c20541d5953674055dd345065aa8c3e0981d7bc4b392927d82c2a1614eb6dc9e"}
{"ts":"2026-05-21T17:30:28.764Z","author":"codex-gpt5","op":"files_add","patch":[{"op":"add","path":"/metadata/files/2","value":{"path":"tests/unit/comments-command.spec.ts","scope":"project","note":"comments metadata and text= regression"}},{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:30:28.764Z"}],"before_hash":"c20541d5953674055dd345065aa8c3e0981d7bc4b392927d82c2a1614eb6dc9e","after_hash":"c4f7a1acd21ac820cdeccbd2176dd7343c4c894c90644e79b566c87ea5c8f2b3"}
{"ts":"2026-05-21T17:30:31.228Z","author":"codex-gpt5","op":"files_add","patch":[{"op":"replace","path":"/metadata/files/2/note","value":"MCP comments compact default regression"},{"op":"replace","path":"/metadata/files/2/path","value":"tests/integration/mcp-dynamic-package-actions.spec.ts"},{"op":"add","path":"/metadata/files/3","value":{"path":"tests/unit/comments-command.spec.ts","scope":"project","note":"comments metadata and text= regression"}},{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:30:31.228Z"}],"before_hash":"c4f7a1acd21ac820cdeccbd2176dd7343c4c894c90644e79b566c87ea5c8f2b3","after_hash":"0be62f63aa660d2824e9681986d8f224a269aa974c0a4539fedb423c63c57507"}
{"ts":"2026-05-21T17:30:33.285Z","author":"codex-gpt5","op":"tests_add","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:30:33.285Z"},{"op":"add","path":"/metadata/tests","value":[{"command":"PM_RUN_TESTS_SKIP_BUILD=1 node scripts/run-tests.mjs test -- tests/unit/comments-command.spec.ts tests/integration/mcp-dynamic-package-actions.spec.ts","scope":"project","timeout_seconds":300}]}],"before_hash":"0be62f63aa660d2824e9681986d8f224a269aa974c0a4539fedb423c63c57507","after_hash":"0924ff2d228c77c42e7822494f9a9dec0814a396b378f00bf7dc764f84aeb537"}
{"ts":"2026-05-21T17:40:17.143Z","author":"codex-gpt5","op":"comment_add","patch":[{"op":"add","path":"/metadata/comments/1","value":{"created_at":"2026-05-21T17:40:17.143Z","author":"codex-gpt5","text":"Evidence 2026-05-21: implemented MCP comments token-efficiency default. pm_comments list calls now default to latest 20 comments with total_count, returned_count, has_more, and limit metadata unless options.limit or options.full is supplied. CLI runComments remains backward compatible and only emits metadata when requested. Also fixed comments --add text=... parsing so comma-containing text values strip the text= prefix. Verification passed: combined focused suite 8 files / 310 tests; package-first dogfood 84 workflows; semantic dogfood 86 workflows; full release gate 96 files / 1406 tests at 100 percent coverage plus required Sentry/telemetry."}},{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:40:17.143Z"}],"before_hash":"0924ff2d228c77c42e7822494f9a9dec0814a396b378f00bf7dc764f84aeb537","after_hash":"b69e2aeaa284d5a32cf5725be6d699077bb50145e95719a0065d7e1e54c08c57"}
{"ts":"2026-05-21T17:41:14.231Z","author":"codex-gpt5","op":"close","patch":[{"op":"remove","path":"/metadata/assignee"},{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:41:14.231Z"},{"op":"replace","path":"/metadata/status","value":"closed"},{"op":"add","path":"/metadata/close_reason","value":"Implemented MCP comments compact default with explicit paging metadata and text= parsing fix; verified by focused tests, temp dogfood, semantic dogfood, and full release gate with required Sentry/telemetry."}],"before_hash":"b69e2aeaa284d5a32cf5725be6d699077bb50145e95719a0065d7e1e54c08c57","after_hash":"12ff3091f025c7ed94dc5a6c0db522ef440ef5d80a4b1bf20fed962c3fb1031d"}
{"ts":"2026-05-21T17:41:16.362Z","author":"codex-gpt5","op":"release","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:41:16.362Z"}],"before_hash":"12ff3091f025c7ed94dc5a6c0db522ef440ef5d80a4b1bf20fed962c3fb1031d","after_hash":"cbb25924c26efafcf8fa69792d40db0ebe23b2f3a9fe10459d2bab6a349d4e67"}
{"ts":"2026-05-21T17:41:37.158Z","author":"codex-gpt5","op":"update_audit","patch":[{"op":"replace","path":"/metadata/updated_at","value":"2026-05-21T17:41:37.158Z"},{"op":"add","path":"/metadata/resolution","value":"MCP pm_comments now defaults list calls to a token-efficient latest-20 page with explicit paging metadata; full history remains available via options.full or caller-supplied limit."},{"op":"add","path":"/metadata/expected_result","value":"Agents can inspect long-lived item comments without overflowing context and can tell when older comments exist."},{"op":"add","path":"/metadata/actual_result","value":"MCP comments returns latest comments plus total_count/returned_count/has_more/limit metadata; text= comment payloads strip the prefix even when the value contains commas."}],"before_hash":"cbb25924c26efafcf8fa69792d40db0ebe23b2f3a9fe10459d2bab6a349d4e67","after_hash":"16961a65e453d0738363257750e3f5534e63498a55f62a5ee01e48cda08a2b2e"}
Loading