CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

braintrust-analyze

Analyze Claude Code sessions via Braintrust

Quality

70%

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 ./.claude/skills/braintrust-analyze/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

75%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 strong, command-first skill body: fully executable commands, scenario-to-command mapping, options and prerequisites, and sample output. Its main weaknesses are redundancy — the Examples section duplicates commands already listed and 'What You'll Learn' echoes the command comments — and the absence of any troubleshooting guidance for the API-key/tracing prerequisites it declares.

Suggestions

Collapse the 'Examples' section: keep the sample --last-session output as the single illustration and drop the re-listed '--detect-loops' and '--weekly-summary' commands, which the Commands section already covers.

Trim 'What You'll Learn', which restates the inline command comments; either fold it into 'When to Use' or cut it in favor of the sample output.

Add one line of failure guidance, e.g. 'If a command errors with an auth/empty result, verify BRAINTRUST_API_KEY in ~/.claude/.env and that braintrust-claude-plugin tracing is enabled.'

DimensionReasoningScore

Conciseness

The Commands section is lean and comment-labeled, but the 'Examples' section re-runs three commands verbatim that were already shown ('--last-session', '--detect-loops', '--weekly-summary') with no new information beyond one sample output — clear duplication that should be trimmed. This fits anchor 3 ('Mostly efficient but includes some unnecessary explanation or could be tightened'); it is not 4 because the duplicated command block is a whole redundant section, not a minor instance.

3 / 5

Actionability

Every command is fully executable and copy-paste ready ('uv run python -m runtime.harness scripts/braintrust_analyze.py --last-session' etc.), each flag is annotated inline, an option (--project NAME) and prerequisites (BRAINTRUST_API_KEY) are stated, and a realistic sample output is included. Matches anchor 5 ('Fully executable; copy-paste ready commands; specific examples cover the common cases').

5 / 5

Workflow Clarity

The 'When to Use' section maps each scenario (retrospective, debugging, weekly review) to the right command, and every command is unambiguous for its single purpose — no destructive or batch operations, so no validation loop is required. It falls just short of anchor 5: there are no checkpoints for the common failure modes (e.g. what to check if BRAINTRUST_API_KEY is missing or tracing was never enabled), which the Requirements section mentions but gives no recovery path for.

4 / 5

Progressive Disclosure

The body is well-organized into clear sections (When to Use, Commands, Options, What You'll Learn, Examples, Requirements) with no nested references and no monolithic inline reference dump. It does not reach anchor 5: the 'What You'll Learn' section partially restates the Commands comments, and the duplicated Examples block is organizationally redundant content that a leaner single-section layout (or a separate examples file) would handle better.

4 / 5

Total

16

/

20

Passed

Description

52%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 is distinct and honest but under-sold: it states a domain and one generic verb, omitting all the concrete capabilities (loop detection, token trends, weekly summaries, session replay) and any 'Use when' trigger guidance. It reads closer to a title than a description, which hurts both completeness and trigger-term quality.

Suggestions

Enumerate the concrete capabilities in the description, e.g. 'Analyze Claude Code sessions via Braintrust: detect repeated-tool loops, replay session action sequences, and track agent, skill, and token usage trends.'

Add an explicit trigger clause, e.g. 'Use when doing a post-task retrospective, debugging why a session failed, reviewing weekly productivity, or investigating token spend.'

Include natural synonyms users would actually say (retrospective, token usage, stuck/loop, session history) so the skill triggers on real phrasing rather than only the technical term 'Braintrust'.

DimensionReasoningScore

Specificity

The description names the domain ('Claude Code sessions', 'Braintrust') but the only action is the generic verb 'Analyze' — no concrete capabilities like loop detection, token trends, or session replay are stated. This matches anchor 2 ('Names the domain but actions are minimal or generic', e.g. 'Processes PDF files'); anchor 3 requires 1-2 concrete actions, which a single generic 'Analyze' does not provide.

2 / 5

Completeness

The 'what' is present but vague ('Analyze... sessions') and the 'when' is entirely absent — no 'Use when...' clause or equivalent trigger guidance, which caps completeness at 3 per the judging guidelines. It is not 2 because the 'what' names a specific, understandable domain rather than being purely vague.

3 / 5

Trigger Term Quality

'Analyze Claude Code sessions via Braintrust' contains relevant keywords ('analyze', 'sessions', 'Claude Code') but misses natural phrases users would say such as 'retrospective', 'token usage', 'loops', 'productivity review', or 'tracing'. This matches anchor 3 ('Some relevant keywords but missing common variations or synonyms'); not 4 because several natural trigger terms present in the body are absent from the description.

3 / 5

Distinctiveness Conflict Risk

The Braintrust-specific framing carves out a clear niche: session analysis via a particular tracing platform. It is unlikely to fire for unrelated skills (PDF work, git commits, document editing). Matches anchor 5 ('Clear niche with distinct triggers; minimal conflict risk').

5 / 5

Total

13

/

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

referenced_paths_exist

Referenced path issues: 11 missing

Warning

Total

15

/

16

Passed

Repository

parcadei/Continuous-Claude-v3

Path

.claude/skills/braintrust-analyze/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.