CtrlK
BlogDocsLog inGet started
Tessl Logo
PluginPublic

ctxd/ctxd-source-context

Use ctxd to inspect tracked source paths and read current context for an agent task. Apply when a user wants answers grounded in changing project sources, needs to refresh indexed context, or wants to verify an edit or removal through ctxd.

Quality

82%

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

Overview
Quality
Evals
Security
Files

Quality

Content

81%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 body is a lean, highly executable guide: concrete CLI commands, an MCP config snippet, and explicit validation/error-recovery checkpoints across every scenario flow. The main gaps are modest — a version pin (0.1.16) that ages, setup detail that could be split into a reference file, and no concrete example MCP call for the primary read workflow.

DimensionReasoningScore

Conciseness

The body is lean, opening with a one-line orientation ('ctxd synchronizes source content into ctxfs') and then command blocks with minimal preamble. Minor trimmable instances remain: the version pin 'The commands above match the published 0.1.16 package' (time-sensitive, not in a deprecated section) and asides like 'A client running elsewhere does not gain access to this machine's store from this configuration.'

4 / 5

Actionability

Provides copy-paste-ready commands throughout ('pip install ctxd', 'ctxd backend set ctxfs', 'ctxd folders add /path/to/source --name Source', 'ctxd files grep "search text" --prefix Source --limit 20') plus a concrete MCP JSON config. It falls short of 5 because the most common path — MCP tool usage ('Begin with `tree` using an empty prefix, a small depth and a bounded limit') — is described in prose without a single concrete example call, and stays above 3 because everything given is executable rather than pseudocode.

4 / 5

Workflow Clarity

Scenario-branched sequences are clearly ordered (connect → inspect → read; sync → re-read → confirm removal) with explicit validation and error-recovery checkpoints: 'inspect `ctxd backend status` and `ctxd folders list` before changing configuration', 'Read the same returned source path again after synchronization', 'If a source was removed, check the parent tree and confirm the old path is unavailable', 'Narrow a query when its result is incomplete', and 'If the installed CLI differs, inspect its help before changing configuration.'

5 / 5

Progressive Disclosure

Four well-scoped sections with no nested references; external links (homepage, source repo) are one level deep and content is appropriately inline for a ~70-line single-purpose skill. Not 5 because the 'Set up ctxd when needed' section inlines install and MCP-client configuration detail that could live in a reference file, keeping SKILL.md closer to a pure overview; not 3 because nothing that clearly belongs in a separate file is inlined at length.

4 / 5

Total

17

/

20

Passed

Description

76%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 clearly names the tool and its actions and includes an explicit, multi-condition 'Apply when' clause, making it complete and largely distinct from other skills. Its main weakness is trigger phrasing that echoes tool terminology rather than the natural words a user would say, which limits keyword coverage.

Suggestions

Add natural synonyms to the trigger clause — e.g. 'when the user wants answers based on current source code, asks to sync or refresh indexed context, or needs to verify a recent edit or removal' — so trigger terms match phrases users would actually say.

Replace tool-centric wording like 'tracked source paths' and 'read current context for an agent task' with user-facing language (e.g. 'registered source folders', 'read up-to-date source content') to broaden natural keyword coverage.

DimensionReasoningScore

Specificity

Names the tool and several concrete actions: 'inspect tracked source paths', 'read current context', 'refresh indexed context', 'verify an edit or removal'. Not 5 because the actions stay at a moderate level of granularity without fuller coverage of what inspection involves; not 3 because it lists more than 1-2 specific actions.

4 / 5

Completeness

Explicitly answers both parts: the what ('Use ctxd to inspect tracked source paths and read current context for an agent task') and an explicit multi-condition when ('Apply when a user wants answers grounded in changing project sources, needs to refresh indexed context, or wants to verify an edit or removal through ctxd'), matching the anchor-5 pattern of concrete trigger phrases.

5 / 5

Trigger Term Quality

Some relevant trigger phrases are present ('answers grounded in changing project sources', 'refresh indexed context', 'verify an edit or removal'), but the wording mirrors tool jargon ('tracked source paths', 'indexed context') and misses natural variations users would actually say, such as 'current source code', 'sync', or 'up to date'.

3 / 5

Distinctiveness Conflict Risk

Anchored on the named tool ctxd with distinct triggers, giving a clear niche. Not 5 because triggers like 'answers grounded in changing project sources' could also fire for generic source-reading or codebase-context skills; not 3 because the ctxd anchoring is strong throughout.

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

metadata_version

'metadata.version' is missing

Warning

Total

15

/

16

Passed

Reviewed

On this page