Analyze repo, tracker, PR, validation, session, and .harness evidence. Use when multi-stage Harness Engineering work needs safe resume routing.
45
47%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./Plugins/harness-engineering/skills/he-reconcile/SKILL.mdCoordinate state, not ceremony. he-reconcile identifies the earliest
incomplete HE stage and preserves the evidence chain that lets the next agent
act immediately. Local AGENTS.md, rules, hooks, command boundaries, and
approval gates outrank this skill.
Use when HE work spans stages or evidence surfaces and needs refresh, resume control, source-prompt coverage, repeated-failure reconstruction, or conflict resolution.
Do not use for a single selected implementation slice, plain code fix, isolated
docs edit, or solved-problem capture. Route direct execution to the active
stage skill and durable learning refresh to he-reinforce.
Goal, repo path, tracker state, specs, plans, PRs, validation, session evidence, source-prompt baseline, repeated-failure evidence, or artifact conflict.
Return mode, stage map, earliest incomplete stage, owner, blockers, next action, retained references, coverage, repeated-failure state, validation, Codex provenance, PR safety trace status, and handoff.
Resolve canonical repo/source and cited evidence before treating it as fact.
Classify side effects before acting. Learning capture is out of scope and must
hand off to he-reinforce.
Start with 2-3 focused surfaces before widening.
.harness evidence.Fail fast. Check routing, stage artifacts, source-prompt coverage, tracker/PR
links, Project Brain freshness, validation evidence, and handoff authority.
Report gates as pass, fail, or blocked. Treat stale tracker, validation,
PR, or artifact evidence as degraded, not closure proof.
When session evidence is cited, check Codex provenance source, redaction status,
proof limits, and PR-safe trace fields; raw local identifiers or transcript/
rollout paths in public text are blockers.
he-reconcile reconstructs state and routes the next stage. Do not collapse
multi-stage work into execution, write user/global config, update external
systems, refresh Project Brain, create learnings, or perform destructive
actions without explicit authority. A route or safe_to_continue status cannot
authorize implementation, tracker mutation, sync/install, destructive cleanup,
If evidence, tracker linkage, source baseline, route, destination, or authority
is missing, stop with blocked_reason and the smallest recovery step. Chat
summaries cannot replace source, tracker, artifact, PR, validation, or trace
evidence. If local proof is strong but tracker state is stale/unavailable, route
to closure evidence recovery instead of declaring done.
Route brainstorm gaps to he-brainstorm, acceptance gaps to he-spec,
strategy gaps to he-plan, implementation to he-work, review to
he-code-review, repeated repair tracking to he-linear-plan/Linear, closure
proof to he-eval-report, and solved-problem capture to he-reinforce.
Structured output: schema_version, mode, stage_map,
earliest_incomplete_stage, active_owner, blockers, next_action,
source_prompt_coverage, repeated_failure_state, blackboard_delta,
retained_references, validation, git_staging_status, staged_paths,
codex_provenance, pr_safety_trace, handoff, and blocked_reason.
.harness/session-evidence/latest.md and it
conflicts with .harness/plan/JSC-246-plan.md, route to the earliest
incomplete stage and mark stale plan evidence as degraded.he-eval-report instead of declaring completion.he-reinforce owns new or refreshed .harness/solutions/** and Project Brain
learning synchronization.Read references/contract.yaml for the full reconcile contract and
references/evals.yaml for validation scenarios. Use shared HE references only
when active: stage context, coding-harness bridge, source-prompt coverage,
plugin-hook capability, UI plan routing, artifact routing, agent-native
compression, pragmatic invariants, and XP operating contract. Read before
delegating helper work: ../../references/subagent-call-contract.md.
Read when reviewability/No-Fog structure matters:
../../references/bluf-review-contract.md.
Read when source-of-truth conflicts or route decisions need diagrams:
../../references/visual-reference-contract.md.
4c78f98
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.