Agent skill

context-anchor

Interviews the team once and writes a single GTM_CONTEXT.md (ICP, positioning, voice, proof, competitors) that every other GTM skill reads, then scores how fresh and complete it is.

activeSelf-containedInstructions only445 words

Filed under Positioning and messaging.

From varunk130/ai-gtm-skill-library · 37 skill entries · 6 · pushed 2026-09-19

What it does when it runs

Interviews the team once and writes a single GTM_CONTEXT.md (ICP, positioning, voice, proof, competitors) that every other GTM skill reads, then scores how fresh and complete it is. Use when: set up gtm context, company context file, shared positioning file, brand context, icp file, onboard the gtm skills, context is stale, refresh positioning.

Automated analysis of the skill and the 0 files bundled beside it. A skill’s own description is written to be selected by an agent, so it describes the job and not the dependencies.

Keys and connectors you must supply: None found.
Hosts it reaches: No third-party host appears in the skill or its bundled files.
Tool permissions it declares: No allowed-tools in the frontmatter. It only issues instructions, so there is nothing to bound.
Actions present in the files: None. Instructions only.

Our read

Computed from the artifact. Nothing here has been run, so none of it is a claim about how well the skill works — only about what it is and what it needs.

Maintenanceactive

Pushed within the last 30 days, with commits in the last 90.

CouplingSelf-contained

Nothing outside the agent is required.

Blast radiusInstructions only

No shell, no file writes, no network calls in the skill or its bundled files.

What it does to your systemsAdvice only

No steps that read external data, write records, send messages or spend credits were found in this skill's instructions.

Sends messages: No step found in the source
Writes records: No step found in the source
Spends credits: No step found in the source
Needs credentials: No key or connector found by pattern match

Read from source, not tested by running. Machine-read from SKILL.md at commit 158f12c with jev-1.13.0; labels that write, send or spend were also checked against the cited line by a second reviewer. How this is read

Substancebrief

Under 600 words, no bundled reference files.

LicenceMIT

Permits redistribution with attribution, per the repository.

Ask about context-anchor

Opens your assistant with this page's verified links already in the prompt.

Is this safe to install?Claude ChatGPT

Adapt it to my stackClaude ChatGPT

What else do I need for it to workClaude ChatGPT

Rather ask a human? Talk to Cheetah

Install it

View source on GitHub ↗

git clone --depth 1 --filter=blob:none --sparse https://github.com/varunk130/ai-gtm-skill-library.git /tmp/ai-gtm-skill-library
git -C /tmp/ai-gtm-skill-library sparse-checkout set "gtm-skills/context-anchor"
mkdir -p ~/.claude/skills/context-anchor
cp -R "/tmp/ai-gtm-skill-library/gtm-skills/context-anchor/." ~/.claude/skills/context-anchor/

Picked up without a restart. A project skill of the same name is shadowed by your personal one. For one repository only, swap ~/.claude/skills for .claude/skills. Claude Code docs ↗

The folder is the same in every client that implements the format — 46 of them — so if yours is not above, only the destination changes.

The skill

Source on GitHub ↗

Reproduced in full from varunk130/ai-gtm-skill-library/blob/158f12c03830dca454576eb1a78b1e4a34ae83e3/gtm-skills/context-anchor/SKILL.md, which is licensed MIT (repository). 445 words, 7 headings.

Context Anchor (CANVAS Framework)

Most GTM skills share one gap: each one re-asks who the buyer is, what the product does, and why it wins, and gets a slightly different answer every time. Context Anchor captures those answers once in a versioned file, then keeps the file honest with a freshness and completeness score so downstream skills never run on stale positioning.

Core Principle

One source of truth, dated and scored. A context file nobody has touched in a quarter quietly spreads outdated positioning, because every skill downstream repeats it.

The CANVAS Framework

Letter	Section	The Question
C	Claim	What do we say we do better, in one sentence a buyer would repeat?
A	Audience	Who buys, who uses, who signs? (ICP firmographics + personas)
N	Need	Which job or pain triggers a purchase right now?
V	Voice	How we sound: words we use, words we never use, reading level
A	Alternatives	What else the buyer considers, including doing nothing
S	Substantiation	Which numbers, logos, and quotes back the claim?

Process

Interview - ask one CANVAS section at a time; accept "unknown" rather than guessing
Draft - write GTM_CONTEXT.md with one H2 per CANVAS section and a last_reviewed date per section
Score completeness - each section scores 0 (missing), 1 (asserted), or 2 (asserted + evidence); completeness = total / 12
Score freshness - per section, freshness = max(0, 1 - days_since_review / 90)
Gate - context health = completeness × average freshness; below 0.6, downstream skills should warn before running
Hand off - list which sections each paired skill reads so owners know what to update

Output

Save to GTM_CONTEXT.md at the repo root, plus outputs/context-anchor-health-[YYYY-MM-DD].md

Artifact	Description
GTM_CONTEXT.md	Six CANVAS sections, each with an owner and `last_reviewed` date
Health table	Completeness (0-2) and freshness (0-1) per section, plus overall health
Refresh list	Sections below 0.5 freshness or missing evidence, with the owner to ping

Tips

Evidence or it didn't happen - a claim without proof scores 1, not 2
Name the do-nothing option - it wins more deals than any competitor
Keep Voice short - ten "use" words and ten "avoid" words beat a style guide nobody reads

Pairs With

position-lock - reads Claim, Alternatives; writes back a sharper Claim
competitive-battlecard - reads Alternatives and Substantiation
product-announcement, demand-engine - read Audience, Voice

Other skills for the same job

Different authors, same problem. Matched on the words in the skill name, across every library in the catalogue except this one.

startup-context by shawnpang · 331
multi-turn-context-retention by louisblythe · 159
context-building by extruct-ai · 107
cargo-context by getcargohq · 19
product-context by taizen-ai · 8
context-building by getbeton · 6
context-hygiene by getcargohq · 6
client-context-onboarding by Ad-Superpowers · 5

Need help setting it up?

This page tells you what context-anchor does and what it needs. Cheetah builds the agent setup it runs inside: data, CRM, sequencing and the guardrails.

Book a call →

The directory stays free. There is nothing gated behind this.