CtrlK
BlogDocsLog inGet started
Tessl Logo

evaluate-sessions

Evaluate the build trail of a PR — read the claude -p sessions the harness ran to build it, find where the project's context (Expert / AGENTS.md / skill / spec) served or failed the agents, then capture the learnings as evals (regression tests over the harness's own skills/context) and context fixes. Use when resolving a STUCK (diagnosis-first), auditing how a converged PR was built, or auditing a /learn memory PR. Human-driven and conversational — the trail-evaluating sibling of /evaluate-pr. Outcomes land on a branch (the PR's, or a fresh capture branch if you'll discard the PR) and reach memory via merge + /learn. Triggers - evaluate-sessions, evaluate sessions, review the build trail, diagnose stuck, audit how this was built, session observability.

65

Quality

77%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./skills/human-loop/evaluate-sessions/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

67%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.

A well-architected forensic-process skill: clear sequenced workflow, concrete executable commands, and clean progressive disclosure into real reference files. The main drag is conciseness — motivational framing and restated constraints in the 'Hard nevers' section add tokens without adding guidance.

Suggestions

Trim the 'Hard nevers' section to only the constraints not already stated in S5/S6/S8/S4, or convert it to a one-line cross-reference, to reduce restatement.

Cut motivational prose ('say it to the human, because it's the point', 'Why this is exciting') or fold it into the relevant step so the body stays procedural.

Make the Step 6 landing validation mechanical where possible (e.g. assert the branch is not main and the eval went RED-then-GREEN) rather than relying on a human confirmation.

DimensionReasoningScore

Conciseness

Mostly efficient and free of generic concept-explanation, but the body restates several ideas (the S9 routing and S4/S5/S6 constraints recur in the 'Hard nevers' section) and carries motivational prose ('say it to the human, because it's the point') that could be tightened.

3 / 5

Actionability

Provides concrete, copy-pasteable commands (resolve-sessions.sh invocation, the detached-checkout git sequence, `gh pr view ... --jq`, `jq` trace filters), with the deeper eval template deferred to references — mostly executable with minor gaps.

4 / 5

Workflow Clarity

Steps 0–7 are clearly sequenced with checkpoints (clean-tree preflight, RED-before/GREEN-after right-reason check, end-on-main return), though some validation is soft ('confirm the human's explicit choice') rather than mechanical.

4 / 5

Progressive Disclosure

Well-structured overview with four real one-level-deep reference files and one script, each clearly signaled with its contents and a hackable-seam note; the philosophy section is dense but load-bearing rather than misfiled detail.

4 / 5

Total

15

/

20

Passed

Description

87%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.

A strong description that names concrete actions, gives explicit 'what' and 'when' guidance with natural trigger terms, and clearly distinguishes itself from the sibling /evaluate-pr skill. The main weakness is mild abstraction in a couple of action verbs.

DimensionReasoningScore

Specificity

Lists multiple concrete actions ('read the claude -p sessions', 'find where the project's context... served or failed', 'capture the learnings as evals... and context fixes'), but the abstract phrasing of 'served or failed the agents' keeps it just below the crispest comprehensive anchor.

4 / 5

Completeness

Explicitly answers both what (read sessions, locate context failures, capture evals and context fixes) and when ('Use when resolving a STUCK..., auditing how a converged PR was built, or auditing a /learn memory PR') with concrete trigger phrases.

5 / 5

Trigger Term Quality

Includes natural trigger phrases a user would say ('evaluate-sessions, evaluate sessions, review the build trail, diagnose stuck, audit how this was built, session observability') with good coverage, though a few synonyms/variations are missing.

4 / 5

Distinctiveness Conflict Risk

Carves a clear niche (build trail of claude -p sessions) and explicitly distinguishes itself from /evaluate-pr ('the trail-evaluating sibling of /evaluate-pr'), giving it distinct triggers with minimal conflict risk.

5 / 5

Total

18

/

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.

Validation16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
tdg-ninja/context-specs-factory-ai
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.