CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

track-intent

Keeps a private record of what the developer and Claude have agreed in this session (requirements in the developer's own words, assumptions, open questions, suggestions and what-ifs that weren't agreed, and what has been built) and reads every new message against it, so a question doesn't become an instruction and a correction doesn't erase other requirements. Knackpack's session-start hook loads it automatically; run it yourself only where hooks are blocked, or to see the agreement, continue earlier work, or stop tracking.

Quality

83%

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

SKILL.md
Quality
Evals
Security

Quality

Content

82%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.

An unusually disciplined instruction-only skill: every line carries a rule, the record template and manual command are copy-paste ready, and the per-message workflow has genuine validation checkpoints. The only meaningful weaknesses are minor over-explanation of familiar distinctions and an implicit — rather than explicitly ordered — workflow spread across sections.

DimensionReasoningScore

Conciseness

The body is dense, rule-per-line prose with no filler and no teaching of generic concepts, so it sits at 'efficient; minor instances of over-explanation that could be trimmed' — e.g., 'signing in (authentication) and permission (authorization)... retrying and being safe to repeat (idempotency)' re-explains distinctions Claude already knows, and the Glossary bullet is very long. It is not 5 because those passages could be tightened, and not 3 because nothing is padded or unnecessary in bulk.

4 / 5

Actionability

For an instruction-only skill the guidance is fully executable: a copy-paste command ('sh "${CLAUDE_PLUGIN_ROOT}/hooks/track-intent.sh" manual ${CLAUDE_SESSION_ID} "${CLAUDE_PROJECT_DIR}"'), a complete record template with an ID scheme and section semantics, worked message examples ('maybe Redis would help', 'managers can see their own organization's orders'), exact paths ('.claude/rules/glossary.md'), and an exact output convention ('Noted:'). It is not 4 because no gaps remain in the executable guidance for this skill type.

5 / 5

Workflow Clarity

The per-message flow is well sequenced (classify the message → ask only when readings diverge materially → update the record → report) with real checkpoints: 'Read the file before each update', 'Update it in the same turn, before you start any work the message approves', 'never write "Noted:" without updating the record', and '"verified" needs a check that ran and passed'. It is not 5 because the sequence must be synthesized across three sections rather than presented as an ordered workflow, and fork/continuation edge cases live in a separate section; it is well above the 3 anchor since checkpoints are explicit, and the destructive/batch cap does not apply to record-keeping.

4 / 5

Progressive Disclosure

There are no bundle files (no references/, scripts/, or assets/), and the single ~90-line SKILL.md is well organized with clear ## sections, everything at one level and no buried or nested references — matching 'good structure; most content is appropriately placed... minor organization gaps'. It is not 5 because the under-50-line simple-skill exception does not apply and some content (the Glossary and Sessions procedures, the record layout) could arguably live in separate reference files; it is not 3 because the structure is good and nothing is badly misplaced.

4 / 5

Total

17

/

20

Passed

Description

78%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 dense, information-rich description that clearly answers both what the skill does and when it should run, with concrete trigger phrases and a distinct niche. Its main deductions are a second-person clause that draws the rubric's specificity penalty and keyword coverage that misses a few natural synonyms users might say.

Suggestions

Rewrite the final clause in third person (e.g., 'Run it manually only where hooks are blocked, or to view the agreement, continue earlier work, or stop tracking') to avoid the second-person specificity penalty.

Add one or two natural synonyms users would actually say — such as 'what we agreed on', 'decisions', or 'keep track of requirements' — to broaden trigger-term coverage.

DimensionReasoningScore

Specificity

The description lists several concrete actions ('Keeps a private record of what the developer and Claude have agreed... requirements... assumptions, open questions... and what has been built', 'reads every new message against it'), which is a 4 ('several specific actions; minor gaps'), but the explicit second-person clause 'run it yourself only where hooks are blocked, or to see the agreement, continue earlier work, or stop tracking' triggers the rubric's 1-point specificity penalty. It is not a 2 because the domain and many concrete actions are explicitly named rather than generic.

3 / 5

Completeness

Both questions are answered explicitly: what ('Keeps a private record... and reads every new message against it, so a question doesn't become an instruction') and when, with concrete trigger phrases ('Knackpack's session-start hook loads it automatically; run it yourself only where hooks are blocked, or to see the agreement, continue earlier work, or stop tracking'). The explicit trigger guidance means the missing-'Use when' cap does not apply, and the 'when' is specific enough to exceed the 4 anchor.

5 / 5

Trigger Term Quality

Natural phrases a user would say are present — 'the agreement', 'continue earlier work', 'stop tracking', 'requirements', 'session' — giving good keyword coverage. It is not 5 because common variations such as 'decisions', 'what we agreed on', or 'remember what we decided' are missing, though it clearly exceeds the 3 anchor's 'some relevant keywords but missing common variations'.

4 / 5

Distinctiveness Conflict Risk

The niche is clear — session agreement/intent tracking with hook integration — and its triggers (agreement record, continue earlier work, stop tracking) are distinct from most skills. It is not 5 because the somewhat abstract 'Keeps a private record' framing could overlap with generic note-taking, summary, or memory-keeping skills; it is well above the 3 anchor's 'could still overlap with similar skills'.

4 / 5

Total

16

/

20

Passed

Validation

93%

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

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository

bezmoradi/knackpack

Path

skills/track-intent/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.