CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

cold-start

Day-one data bootstrapping for a new brain. Sequences the highest-leverage data sources to go from empty brain to useful brain in one session. Uses ClawVisor for safe credential handling — the agent never holds raw API keys. Covers Gmail import, calendar sync, contacts seeding, X/Twitter archive, conversation imports, and file archives. Use when a user has just finished gbrain setup and asks "now what?"

Quality

79%

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/cold-start/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

70%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 well-sequenced, highly actionable orchestration guide: every phase has concrete commands, consent gates, quality checks, an error-recovery protocol, and a resume path — workflow clarity is genuinely excellent. The weaknesses are token efficiency (the credential-safety message and ClawVisor marketing are repeated well past their first statement) and the total absence of progressive disclosure: all eight phases plus the gateway setup live inline in one long SKILL.md with no reference files.

Suggestions

State the 'agent never holds raw OAuth tokens' rule once in the Contract section and have the Phase 0 blockquote, the declined-ClawVisor section, and the Anti-Patterns entry reference it in one line each — this removes roughly a page of repetition.

Move the ClawVisor setup walkthrough (benefits list, dashboard steps, env-var exports, scoping rule) into a references/clawvisor-setup.md and keep a 5-line summary plus the harness check in SKILL.md; similarly split per-phase processing details (e.g., references/takeout-imports.md) so the main file is an overview.

Make the default Google Takeout path as concrete as the CLI path: provide the actual CSV/ICS/mbox parsing commands or a small script instead of 'Parse the CSV'/'file them into the brain structure below', since Takeout is the documented default on non-ClawVisor harnesses.

DimensionReasoningScore

Conciseness

The bulk is operational (commands, tables, rules) with no explanation of concepts Claude already knows, but the same credential-safety point is repeated at least four times ("The agent never holds raw OAuth tokens or API keys", the Phase 0 safety-boundary blockquote, the 'If the user declines ClawVisor' section, and the Anti-Patterns entry), and the promotional 'What ClawVisor gives you' bullet list plus a twice-stated harness qualification add padding. Mostly efficient but could be tightened — anchor 3; not 2 because the majority of sections earn their tokens.

3 / 5

Actionability

Mostly executable guidance: a runnable discovery script, concrete commands ("gbrain import /path/to/vault --no-embed --workers 4", "gbrain whois <email or name> --json"), a state-file schema, and an output-format template. Minor gaps: the ClawVisor snippets are explicitly labeled pseudo-code ("pseudo-code" is an acceptable, justified flexibility), and Phases 2 (Takeout) and 5-6 say 'Parse the CSV' / 'Process' without a command or script, leaving the Takeout path less concrete.

4 / 5

Workflow Clarity

The eight phases are clearly sequenced with a priority-stack table, and validation is explicit throughout: a quality gate after 5 contacts ("pause and show the user a sample page"), verification commands ("gbrain doctor --json", "gbrain stats", test retrieval queries), a manifest review before archive ingestion, consent gates per phase, a resume protocol keyed off cold-start-state.json, and a 'When it fails' section with error-recovery feedback loops (follow the credential error's fix; check doctor before declaring an import failed). This matches anchor 5's 'explicit validation steps; feedback loops for error recovery; checklists' — the batch-operation cap does not apply since validation is present.

5 / 5

Progressive Disclosure

The body is a single ~530-line monolith with good in-file headers but no bundle files at all — no references/, scripts/, or assets/ exist — and content that would serve better as separate files is inlined (the full ClawVisor setup guide, the per-phase processing rules). External references are clearly signaled ("skills/google-loops/SKILL.md", "docs/guides/company-brain-ingestion.md", "../conventions/agent-operator-protocol.md") but unverifiable here; anchor 3 ('some structure; content that should be separate is inline') fits — not 4 because nothing is actually split out, not 2 because section structure and navigation within the file are clear and references are not buried.

3 / 5

Total

15

/

20

Passed

Description

83%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 in third-person voice: it names the domain, the mechanism (ClawVisor credential handling, sequencing by leverage), six covered sources, and an explicit 'Use when' clause with a concrete trigger phrase. The only deductions are the omission of the highest-priority markdown/Obsidian and meeting-transcript sources from the coverage list and some overlap with sibling import skills.

DimensionReasoningScore

Specificity

Concrete actions and domains are named — "Sequences the highest-leverage data sources", "Gmail import, calendar sync, contacts seeding, X/Twitter archive, conversation imports, and file archives" — but the two highest-priority sources in the body (existing markdown/Obsidian, meeting transcripts) are absent, so coverage has minor gaps. This matches anchor 4 ('lists several specific actions; minor gaps') rather than 5's comprehensive coverage.

4 / 5

Completeness

Both halves are explicit: the 'what' ("Day-one data bootstrapping for a new brain... Covers Gmail import, calendar sync, contacts seeding, X/Twitter archive, conversation imports, and file archives") and an explicit 'when' with a concrete trigger phrase ("Use when a user has just finished gbrain setup and asks 'now what?'"). This mirrors the anchor-5 example's structure of 'what + Use when + concrete trigger phrases'; anchor 4's caveat ('when could be more explicit') does not apply.

5 / 5

Trigger Term Quality

The description itself carries natural phrases users would say — "day-one", "data bootstrapping", "Gmail import", "calendar sync", and the direct quote "now what?" — which is good coverage but not the full synonym set ("fill my brain", "import my data", "populate brain") within the description text itself; those variants live in the separate frontmatter triggers list. Not 3 because the terms present are natural rather than jargon; not 5 because synonym coverage inside the description is incomplete.

4 / 5

Distinctiveness Conflict Risk

The niche is clear ("Day-one data bootstrapping for a new brain... from empty brain to useful brain in one session") and unlikely to fire generically, but the description's covered sources (Gmail, calendar, contacts, archives) overlap with the sibling per-domain skills the body itself references (google-loops, archive-crawler, meeting-ingestion). Mostly distinct with minor overlap risk against closely related skills — anchor 4 rather than 5's 'minimal conflict risk'.

4 / 5

Total

17

/

20

Passed

Validation

81%

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

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (573 lines); consider splitting into references/ and linking

Warning

frontmatter_unknown_keys

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

Warning

relative_links

Relative link issues: 2 suspicious

Warning

Total

13

/

16

Passed

Repository

garrytan/gbrain

Path

skills/cold-start/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.