private-company-research

Deep research framework for pre-IPO / private companies (Ant Group, SpaceX, Stripe, ByteDance...). Six analyst lenses — business model, financial forensics, competitive landscape, risk & governance, tech & IP, alternative-data signals — run in parallel via run_swarm, then cross-validated for signal consistency before any verdict. Built around the core challenge of private-company work: information is scarce, so every data point carries a confidence label (high / medium / low), inference is shown separately from fact, and 'I don't know' is a valid output. Outputs a fair-value range, exit-path analysis, and an information-gap map. Use for any unlisted company where you need to judge what the business is actually worth.

Quality

90%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

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.

The body is an unusually dense, highly actionable research framework — specific sources, numeric parameters, a mandatory cross-validation gate, and a fixed output contract. Its main costs are redundancy between Key Principles and earlier sections, a missing error-recovery loop around the report_audit gate, and a monolithic single-file structure where per-lens detail could be externalized.

Suggestions

Deduplicate 'Key Principles' by cutting items 3, 4, 7, 9, and 10, which restate the Cross-Validation, Data Labeling, and AI-bias sections verbatim in intent.

Add one error-recovery step after the 'report_audit' quality gate — e.g., what class of numeric/labeling failures to fix and that the audit must be re-run until it passes.

Move the six per-lens checklists (and optionally the 14-section report template) into references/ files (e.g., references/lenses.md, references/report-template.md) so SKILL.md stays a navigable overview.

Dimension	Reasoning	Score
Conciseness	The body is telegraphic and dense with no padding of concepts Claude already knows — every lens line carries methodology specifics (metrics, scales, source names). It falls short of a 5 because 'Key Principles' items 3, 4, 7, 9, and 10 restate the Cross-Validation, Data Labeling, and bias self-check sections earlier in the file.	4 / 5
Actionability	For an instruction-only skill the guidance is fully concrete: named data sources (QuestMobile, Sensor Tower, Google Patents, 天眼查/企查查, LatePost/The Information), exact parameters ('20-40% discount', 'liquidity discount 20-30%', '3-5' comparables, '5-7' bull/bear points), a fixed 14-section report template, a concrete save path 'reports/{company}/{company}-private-{YYYYMMDD}.md', and a 'report_audit' quality gate. Not 4: no key execution details are missing.	5 / 5
Workflow Clarity	The sequence is clear — six lenses in parallel (or sequential), then a 'Cross-Validation (team-lead, mandatory)' gate with a numbered 4-step checklist, then a fixed report structure, then save and 'Run report_audit on the numbers as a quality gate'. Not 5: there is no error-recovery instruction for what to fix and re-run when the audit gate fails; not 3: validation checkpoints are explicit and prominent, not merely implicit.	4 / 5
Progressive Disclosure	No bundle files exist (references/, scripts/, assets/ are all absent), and the single SKILL.md is well organized with clear section headers, a role table, and a one-level-deep structure with no nested references. Not 5: at ~160 lines, the six per-lens checklists and the 14-section report template are candidates for references/ files to keep the overview lean; not 3: organization is strong and nothing is buried.	4 / 5
	Total	17 / 20 Passed

Description

96%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: it enumerates concrete capabilities, names its deliverables, covers natural trigger synonyms, and closes with an explicit 'Use for' clause. The only weakness is that leading with 'Deep research framework' creates slight overlap risk with generic research skills.

Dimension	Reasoning	Score
Specificity	The description lists multiple concrete actions and deliverables — 'Six analyst lenses — business model, financial forensics, competitive landscape, risk & governance, tech & IP, alternative-data signals — run in parallel', 'cross-validated for signal consistency', 'Outputs a fair-value range, exit-path analysis, and an information-gap map' — with comprehensive coverage and no generic filler.	5 / 5
Completeness	It explicitly answers both questions: 'what' via the six lenses, cross-validation step, and named outputs, and 'when' via the explicit trigger clause 'Use for any unlisted company where you need to judge what the business is actually worth.' Third-person voice throughout, so no voice penalty applies.	5 / 5
Trigger Term Quality	Natural user phrasing is well covered with synonyms — 'pre-IPO / private companies', 'unlisted company', 'judge what the business is actually worth' — plus concrete company-name triggers (Ant Group, SpaceX, Stripe, ByteDance) a user would actually say. Not 4: unlike the 'a few natural terms missing' anchor, both the EN synonyms and concrete instance triggers are present.	5 / 5
Distinctiveness Conflict Risk	The niche (private-company valuation research) is clear and the 'Use for' clause scopes triggers tightly to unlisted companies, but the opening phrase 'Deep research framework' invites minor overlap with general-purpose deep-research skills — fits the 'mostly distinct; minor overlap risk' anchor rather than the minimal-conflict anchor.	4 / 5
	Total	19 / 20 Passed

Validation

93%

Warnings & errors only

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

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: HKUDS/Vibe-Trading
Path: agent/src/skills/private-company-research/SKILL.md
Commit: 251b094

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

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.