CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

he-reconcile

Analyze repo, tracker, PR, validation, session, and .harness evidence. Use when multi-stage Harness Engineering work needs safe resume routing.

Quality

68%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./Plugins/harness-engineering/skills/he-reconcile/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

57%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The skill body is token-lean and structurally well-sectioned with explicit validation gates, but it is compressed past the point of self-sufficiency: the procedure truncates mid-step, safety boundaries end mid-sentence, and over half of its cited reference files are missing from the bundle. As shipped, a reader cannot follow the core workflow without hitting broken links or undefined jargon.

Suggestions

Complete the truncated text: finish Procedure step 4 (it ends at "repo drift" without its continuation) and the Safety Boundaries sentence that dangles at "destructive cleanup," so the inline sequence is coherent.

Fix or inline the six missing ../../references/*.md citations (stage-arc-boundary-contract, subagent-call-contract, bluf-review-contract, visual-reference-contract, deferred-context-index, closure-mutation-contract) — more than half of the referenced files are unreachable in the current bundle.

De-duplicate the repeated content: the context-disposition policy sentence appears verbatim in both Gotchas and References, and the hot-path-folded-context.md pointer is repeated in four sections — keep one of each.

DimensionReasoningScore

Conciseness

The body is telegraphic and assumes Claude's competence with no concept explanations, but tokens are wasted: the context-disposition policy sentence appears verbatim in both Gotchas and References, and the "See references/hot-path-folded-context.md" pointer is repeated in four sections. This is 'Efficient; minor instances ... that could be trimmed' rather than anchor 5's 'every token earns its place'.

4 / 5

Actionability

There is genuine concrete material — output field names (stage_map, earliest_incomplete_stage, blocked_reason), explicit status vocabularies (pass/fail/blocked; found/not_found/blocked/not_applicable), and example paths like .harness/session-evidence/latest.md — but the core procedure steps are abstract directives ("Resolve only enough context", "apply source-prompt coverage") whose executable detail is deferred to reference files, and undefined jargon (Project Brain, stage arc boundary) is used without inline definition. This fits 'Some concrete guidance but incomplete ... missing key details', not 4's 'concrete ... with minor gaps'.

3 / 5

Workflow Clarity

A numbered Procedure exists and the Validation section is explicit (fail fast, gate statuses, degraded-evidence rules), but Procedure step 4 truncates mid-sentence at "repo drift" and Safety Boundaries dangles at "destructive cleanup," — the visible sequence is incomplete and the continuation (steps 5-11) lives only in a folded reference. The sequence is present but has real gaps, matching anchor 3 rather than 4's 'clear sequence with most checkpoints present; minor validation gaps'.

3 / 5

Progressive Disclosure

Sections are well-organized and references are one level deep, but scored against the actual bundle: six of the nine distinct referenced files (../../references/stage-arc-boundary-contract.md, subagent-call-contract.md, bluf-review-contract.md, visual-reference-contract.md, deferred-context-index.md, closure-mutation-contract.md) do not exist anywhere in the bundle, so the Stage Arc Boundary and other sections point readers at unreachable files. Combined with the duplicated policy line, this fits anchor 3 ('structure but could be better organized; references present but ...'), not 4's 'minor organization gaps'.

3 / 5

Total

13

/

20

Passed

Description

62%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description has an explicit 'Use when' trigger and clearly named evidence sources, but the capability is expressed as a single generic verb over inputs rather than concrete outcomes. It is serviceable but reads as compressed jargon rather than a complete statement of what the skill does.

Suggestions

State the concrete outcome in the what-clause, e.g. "identifies the earliest incomplete Harness Engineering stage and routes work to the owning stage skill" instead of the generic "Analyze ... evidence".

Add natural user phrasings such as "continue interrupted work", "which stage to resume", or "handoff" alongside the domain jargon so the trigger matches what users actually say.

Briefly distinguish from sibling HE skills (e.g. mention routing vs. execution vs. learning capture) to reduce overlap risk with he-plan/he-work descriptions.

DimensionReasoningScore

Specificity

"Analyze repo, tracker, PR, validation, session, and .harness evidence" names the domain and input surfaces but offers only one generic action ("Analyze") plus "safe resume routing"; it never states what the analysis produces. This matches the anchor 'Names domain and 1-2 concrete actions, but not comprehensive' — not 4, which requires several distinct specific actions.

3 / 5

Completeness

Both parts are explicit: the what ("Analyze repo, tracker, PR, validation, session, and .harness evidence") and the when ("Use when multi-stage Harness Engineering work needs safe resume routing"). Not 5 because the what stops at inputs — the actual outcome (identifying the earliest incomplete stage and routing to the owning skill) is never stated.

4 / 5

Trigger Term Quality

Terms like "repo", "PR", "validation", and "resume" are natural for the domain, but ".harness evidence" and "Harness Engineering" are internal jargon, and common user phrasings ("continue interrupted work", "which stage is incomplete", "handoff") are absent. This is 'Some relevant keywords but missing common variations or synonyms', below 4's 'a few natural terms missing'.

3 / 5

Distinctiveness Conflict Risk

"multi-stage" and "safe resume routing" carve a distinct niche versus single-stage execution skills, giving 'minor overlap risk with closely related skills'. Not 5 because the description does not disambiguate against sibling Harness Engineering skills (e.g. he-plan, he-work) that share the same evidence vocabulary.

4 / 5

Total

14

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

jscraik/Agent-Skills

Path

Plugins/harness-engineering/skills/he-reconcile/SKILL.md

Reviewed

Table of Contents

Is this your skill?

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.