You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
seconder specialist (NEW, package tier). The fused post-writer gate from canon docs/design/chain-templates.md §2.3 — collapses the old split between scope/compliance (reviewer phase-1) and code-quality smell (code-sanity) into one READ_ONLY dispatch (openai-codex/gpt-5.4-mini) emitting a structured dual-verdict JSON: scope_verdict + scope_findings + quality_verdict + quality_findings + overall_verdict. The chain reducer reads overall_verdict to advance or route back to the writer; the reviewer reads the dimension-tagged findings. Replaces code-sanity, which is removed this release (see Removed) (unitAI-4e194, unitAI-wz2ag, unitAI-321ir).
test-engineer specialist (NEW, package tier). Post-implementation behavioral-test author from the actual diff (canon §2.5). MEDIUM, openai-codex/gpt-5.5, requires_worktree. Produces tests + fixtures + smoke/E2E harnesses + telemetry assertions and emits exact test-runner commands via a structured schema (status, files_changed, coverage_map, smoke_e2e_commands, telemetry_assertions, test_runner_commands, known_deferred_paths, source_bug_suspicions). Ambidextrous role (§3.16): the same spec is the primary writer in test-only chains and the secondary writer in code-with-tests chains — the system prompt is mode-agnostic and the position arrives via the dispatch-time mandate. Forbidden from patching production source by default (unitAI-sfwe1, unitAI-sfwe1.1).
Two NEW chain-template formulas — code-with-tests (§3.14) and test-only (§3.15).code-with-tests: dual-writer production chain (executor writes the diff, test-engineer writes tests against it) at high/critical scrutiny. test-only: single-writer chain when scope is test-paths only (test-engineer as primary writer). Both carry the ambidextrous test-engineer mandate in the step description (pre-substrate position-injection mechanism, §3.16) (unitAI-f9kku).
Seconder dual-verdict eval + QA-routing eval..specialists/evals/seconder/ — a reproducible static eval with three fixtures (wrong-scope → scope_verdict FAIL, bad-quality → quality_verdict FAIL, clean → overall_verdict PASS) each carrying an expected-verdict.json, plus an operator-run run.sh and a token-cost note (unitAI-o7j1a). config/skills/using-specialists-v3/evals/ gained four QA-routing eval cases (test-engineer primary vs secondary writer, test-runner owner-routing, reviewer-consumes-QA-evidence) + a passing vitest harness (unitAI-sfwe1.5).
transcriber specialist (NEW, package tier). Promotes the documentation-grade YouTube transcriber prompt to the shipped package catalog at config/specialists/transcriber.specialist.json. v1.6.0 uses openai-codex/gpt-5.3-codex, title-derived transcript/analysis filenames, narrow subtitle language extraction to avoid YouTube 429 fanout, immediate section-by-section writes, dense technical DETAILED SECTION ANALYSIS, TECHNICAL EXTRACTION TABLES, and a coverage/quality audit to prevent shallow “2-line per 5 minutes” outputs (unitAI-jfw26).
sp log runtime/provenance stream — new operator-facing log command for specialist runtime debugging. It reads observability.db, shows dispatch/control/status/error/auto-commit provenance separate from sp feed, supports --json NDJSON for full payloads, --follow, --since, --limit, job/bead/specialist/node filters, and --all-events for raw feed-like internals (unitAI-gqpvw, unitAI-vfqgq).
sp log parent-directory/global mode — when run outside a repo root with no local specialists DB, sp log discovers immediate child repos containing .specialists/db/observability.db and aggregates their runtime rows as one global log; --repo <name> narrows output to a single child repo (unitAI-v5xfu).
obligations-scanner specialist (NEW) — READ_ONLY, cheap (openai-codex/gpt-5.4-mini, bare: true, ~30s target) pre-review marker scan. Scans executor/debugger diffs for newly-introduced TODO/FIXME/HACK/XXX/TEMP/WIP/NOTE(release) markers in production code. Distinguishes production vs test/fixture surfaces. Recognizes structured // TODO(<bead-id>): reason format and treats it as TRACKED when the linked bead is open. Verdict: CLEAN | OBLIGATIONS_FOUND | BLOCKED with a JSON output_schema the reviewer consumes directly. Iron-style obligations tracking (unitAI-kglvm.3).
docs/design/iron-review-hardening.html — design doc visualizing the new pipeline (SCRUTINY taxonomy, old-vs-new chain flow, per-specialist changes, git-state precondition, manual execution plan). Mirrored to ~/second-mind/1-projects/Mercury/ for sync (unitAI-fpwbr, unitAI-1n56e, unitAI-ejdi1).
service-skills-sync specialist (NEW, package tier). Promotes the Service Skills Librarian (previously a market-data user-tier override) to a shipped package specialist at config/specialists/service-skills-sync.specialist.json. MEDIUM, openai-codex/gpt-5.4-mini; keeps per-service expert-persona SKILL.md docs in sync with code drift using gitnexus (detect_changes/impact/context) + Serena, gated by a drift_detector.py pre-scan. The per-service knowledge layer the future devops agent reads (DevOps PRD §7.1) (unitAI-g8zr3).
researcher specialist v1.2.0 → v1.3.0 — general-web pipeline (Mode 4). Adds a fourth research mode closing the web-research gap (previously the researcher reached library docs/repos/code/social but had no general web search or arbitrary-URL read): ddgs (DuckDuckGo search CLI, no API key — uv tool install ddgs) discovers authoritative URLs, then agent-browser (native Rust CLI + Chrome daemon — npm i -g agent-browser) reads any URL including JS-rendered pages. Documented in prompt.system Mode 4 + config/mandatory-rules/research-tool-routing.md. capabilities.external_commands deliberately left empty — it is a hard pre-run gate (runner.ts validateBeforeRun throws on a missing PATH binary), so declaring these heavy tools would break the shipped researcher in projects without them; documented as available-on-demand with install hints instead (unitAI-qgvld).
notes_mode specialist field + markdown-native 3-state handoff. New top-level notes_mode enum (full-trail default | final-only) controls how each turn's handoff is persisted to BOTH the input bead notes and output_file. The supervisor renders a markdown-native 3-state handoff — ### <specialist> · <model> · [turn N · WAITING] trail blocks plus a canonical ## <specialist> · <model> · [FINAL · DONE] block — with the specialist's output verbatim, a single italic metadata footer (empty/zero/unknown fields omitted), and a provider-prefix-stripped model string; no divider rules or emoji. One shared content source feeds bead notes, output_file, and sp result. final-only persists only the canonical FINAL block and overwrites output_file, for non-coding/chained pipelines where the next specialist reads the previous one's note or file as input (unitAI-10y07, unitAI-yiazs).
Changed
test-runner specialist — upgraded to the QA failure-routing contract (canon §2.5). Now prefers exact commands from test-engineer/orchestrator and falls back to manifest-detected runners only when none are supplied (clearly labeled as fallback). Classifies every failure by owner — test_engineer (test/fixture/harness wrong, or new untested feature), debugger_or_executor (missing telemetry / source behavior regression), infrastructure, pre_existing — and never writes tests or patches source (LOW). Backed by config/mandatory-rules/test-runner-execution-scope.md (unitAI-sfwe1.2).
reviewer specialist — refactored to phase-2-only (seconder fusion, canon §2.3). The phase-1 compliance/scope check now lives in seconder's scope_verdict; the reviewer keeps only phase-2 (adversarial deep code-quality audit + machine-readable Release Checklist + ddiff re-review on PARTIAL) and treats a seconder PASS as the upstream scope gate. Two-phase framing removed from the prompt (unitAI-4e194, unitAI-sowpa).
13 chain-template formulas rewired for the canonical QA pipeline. Every production-diff template (code-standard, code-with-advisors, debug, security-deep, restitch) now wires writer → seconder → test-engineer → test-runner → [security-auditor if sensitive] → obligations-scanner → reviewer (canon §2.1) — code-sanity renamed to seconder, test-engineer + test-runner inserted. README overlay table + roadmap Opp 14/15 status updated (unitAI-f9kku).
using-specialists-v3 skill — canonical seconder-fusion pipeline. SKILL.md now teaches writer → seconder → test-engineer → test-runner → [security] → obligations → reviewer, the QA failure-routing matrix (§2.5), and SCRUTINY reframed as a chain property that modulates structure, not quality (§2.2) — including the none tier for read-only chains and the required-at-creation rule. seconder replaces the code-sanity seconder slot; reviewer documented as phase-2-only (unitAI-096re, folds unitAI-sfwe1.3).
sp log human output is leaner and calmer. Default output now hides agent-internal turn/tool/text/thinking/token rows already covered by sp feed, keeps runtime-owned rows only, collapses repo/path/branch/worktree metadata into one compact worktree=<repo>/<worktree> field, uses a restrained professional color palette (dim metadata, plain job ids, bold specialist names, color-coded status=<state>, green/yellow/red/cyan only for semantic state), and collapses adjacent duplicate display rows caused by duplicated runtime events while preserving full payloads in --json (unitAI-vfqgq, unitAI-npjlq, unitAI-f5k0p).
reviewer specialist — Iron-inspired prompt overhaul. Five new system-prompt sections, additions only (existing source-of-truth priority and AUTHORITATIVE REVIEW CONTEXT preserved verbatim) (unitAI-kglvm.1):
SCRUTINY tier behavior (low | medium | high | critical) — reads field from bead contract; defaults to medium; tiers reviewer depth from seconder-only spot-check (low) through file-by-file sign-off with mandatory gitnexus_impact (high) to required second-opinion (critical).
Scrutiny auto-escalation — surface-pattern floor table raises level regardless of bead's stated SCRUTINY when diff touches auth/*, **/credentials*, **/token* (→ high), config/specialists/*.json (→ high), src/specialist/{runner,schema}.ts (→ high), **/*.lock (→ medium + security-auditor required), migrations/** (→ high), src/permissions/* / hooks/** (→ critical). Author's level is a floor, not a ceiling.
Re-review after PARTIAL (Ddiff mode) — when re-reviewing a fixed PARTIAL, scope to delta since prior verdict, carry forward prior approvals, audit only newly-touched files/symbols.
Obligations scan — consumes obligations-scanner JSON output if present, else scans diff inline; production markers → PARTIAL unless accepted via bead NON_GOALS or structured // TODO(<bead-id>): reference; test/fixture markers noted but not blocking.
Release Checklist (REQUIRED) — machine-readable block appended to every verdict for future sp merge enforcement.
executor and debugger specialists — Obligations discipline. New system-prompt section instructs both codegen specialists to avoid introducing in-code obligation markers in production paths by default; if work is genuinely deferred, file a follow-up bead via bd create --deps discovered-from:<current>; if a marker is truly needed at a code site, use structured form // TODO(<follow-up-bead-id>): <reason> where the linked bead is open and listed in current bead's NON_GOALS. Prevents PARTIAL fix-loops from the new obligations-scanner gate. Test/fixture paths exempt (unitAI-kglvm.4).
using-specialists-v3 skill: v3.4 → v3.5 (Iron-style orchestration). Substantial restructure aligned with the above specialist changes (unitAI-kglvm.5):
"Advisory Passes" section reframed as three mandatory gates: Seconder Gate (code-sanity), Security Gate (security-auditor on sensitive surfaces), Obligations Gate (obligations-scanner). Skip rules tightened.
NEW SCRUTINY taxonomy section: tier behavior + auto-escalation surface table. SCRUTINY field added to task/epic, executor, reviewer bead contract templates.
NEW Git State Precondition section: four-check pre-flight (working tree clean, HEAD contains prior chain commits, no orphaned worktrees, in-sync integration branch) required before dispatching any chain that depends on prior chain output. Strictness-by-scenario table.
Rule #9 INVERTED: manual git workflow is now canonical; sp merge and sp epic merge are PROHIBITED (known broken, awaiting separate rework epic). Cherry-Pick Playbook promoted to canonical multi-chain merge path. sp finalize removed from documented orchestrator workflow.
Rule #13 exception clause added for epics that restructure the specialists themselves (operator-authorized manual-orchestrator-direct work).
obligations-scanner row added to Choosing The Specialist table. parallel-review marked deprecated.
Escalation Matrix and Failure Recovery tables rewritten: sp-merge rows replaced with git-workflow recovery patterns (stale .git/index.lock, info/exclude vs tracked beads file, FF-via-git update-ref when checkout blocked).
CLAUDE.md "Common gotchas" section rewritten to match the new canonical: manual merge, explicit sp stop for keep-alive cleanup, Iron-style gates mandatory, Git State Precondition, bd auto-export churn handling, package-tier specialist edits via direct JSON.
bd auto-export pain fix.bd config set export.git-add false disables per-write auto-staging of .beads/issues.jsonl (silent mid-work; no checkout aborts; no .git/index.lock races). Paired with a custom block added to .git/hooks/pre-commit AFTER bd's managed markers — runs git add -f .beads/issues.jsonl so commits naturally include the fresh JSONL snapshot via the existing pre-commit hook chain. Eliminates the runaway chore(beads): export state commits that plagued every multi-bd-op session. Verified end-to-end in this repo (commits 63ac83f6, 4c1f19a5, 1e014f33) (unitAI-mg18o).
output_file decoupled from SPECIALISTS_JOB_FILE_OUTPUT. A specialist that sets output_file now always writes its full result — foreground and --background (tmux) — independent of the env flag, which now only gates the debug file-mirrors (events.jsonl / status.json / result.txt). Previously --background (tmux) runs silently dropped output_file because the env var did not propagate into the tmux session. The single-writer invariant is preserved (the supervisor owns the file in supervised runs; suppressRunnerFileOutput still skips the runner write), and .specialists/*-result.md was added to .gitignore since specs with output_file now always write. output_file, notes_mode, and the handoff envelope are documented in docs/authoring.md and the specialists-creator skill (unitAI-f58ma, unitAI-g8rqg).
Removed
code-sanity specialist removed — superseded by seconder. The Iron seconder gate (briefly promoted as code-sanity mid-cycle, unitAI-kglvm.2) is replaced by the new seconder specialist, which fuses its code-quality smell pass with the reviewer's old phase-1 scope check into one dual-verdict gate (canon §2.3). config/specialists/code-sanity.specialist.json deleted; all operational references across the v3/auto/v2 skills, reviewer.specialist.json, chain-template formula prose, and docs/specialists-catalog.md renamed to seconder. Two historical lineage notes preserved (seconder.specialist.json's absorbed-mandate section + the v3 SKILL §2.3 fusion explanation) (unitAI-321ir, unitAI-4e194).
sp merge / sp epic merge / sp finalize removed from documented orchestrator workflow in using-specialists-v3. Commands still exist in the sp binary (no source-code removal) but the skill explicitly prohibits their use pending a separate rework epic. Operators reaching for them should use the documented manual git workflow instead.
Changed (prior)
All 17 package-shipped specialists in config/specialists/ now declare the v3.16.0 schema additions explicitly: execution.bare: false and prompt.system_prompt_mode: "append". Values match the previous absent-field defaults — pure-mechanical, zero behavior change — but every shipped spec is now self-documenting at the schema level instead of relying on per-runner legacy fallbacks. bare.specialist.json retains its explicit bare: true + replace (unitAI-51r2w).
Fixed
sp stop / sp resume control-plane actions now treat observability writes as best-effort: status/control telemetry failures no longer prevent SIGTERM delivery or falsely report a delivered resume as a steer-pipe write failure (unitAI-dkhi3).
OSV scan now resolves GHSA-q8mj-m7cp-5q26 by overriding all qs lockfile entries to 6.15.2; bun.lock no longer contains vulnerable qs@6.15.1 entries (unitAI-dkhi3).
Supervisor status reads now reconcile dead starting/running specialist jobs to terminal error with a run_complete(ERROR) event, so reviewer crashes during heavy bash validation no longer leave sp ps/sp result stuck on stale running rows (unitAI-6x6p6, unitAI-uzyut).
Per-turn handoff notes now append instead of replace.appendBeadNote called bd update --notes (whole-field replace), so each per-turn specialist handoff clobbered the previous one — multi-turn jobs left only the last (often empty) note on the bead, recoverable only from observability.db. Switched to bd update --append-notes and exported formatBeadNotes. (The appended handoff format was subsequently finalized to the markdown-native 3-state form — see the notes_mode entry under Added.) Tests in bead-notes.test.ts + new supervisor-bead-notes.test.ts (sibling, since supervisor.test.ts is excluded from the default run) (unitAI-sx5qk).
[FINAL · DONE] handoff block now emitted on sp stop for keep-alive jobs. The canonical FINAL block was silently skipped on the dominant keep-alive→sp stop path because src/specialist/control.ts constructed the Supervisor without a beadsClient, so finalizeWaitingJob's bead_id && beadsClient guard never fired; stopJob also never invoked finalizeWaitingJob for waiting jobs. Both fixed, and keep-alive turn summaries skip the duplicate non-final done write, so a keep-alive run yields one [turn N · WAITING] per turn plus one [FINAL · DONE] at stop (unitAI-mis38).
