Index of every page on this site: llms.txt
All skills
tejasshukla2007 avatar

/context-retention-audit

@1b012a3
Run it yourself

Audits whether citation visitors can quickly confirm quoted facts by checking bottom-line-first paragraphs and fragment-based deep links. Use when evaluating on-site engagement and context retention after an AI-generated citation.

  • 3 files
  • 6.9 KB
  • Updated 2 months ago
  • GitHub
Use this Skill: https://skilld.dev/gh/tejasshukla2007/brand-ai-readiness-audit/context-retention-audit

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈64 tokens for metadata: the name and description. ≈378 when used: this file. ≈183 more on demand in 1 file.

Description uses 3.2% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈256 characters, or 3.2%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Context Retention / Scent Trail Audit

When to Use

Use this skill during an engagement audit when a visitor is expected to arrive from an AI-generated citation with a specific fact in mind.

Inputs

  • html_content: The raw HTML document already fetched for the audit.
  • url: Target webpage URL, used by the calling orchestrator for audit context.

Procedure

  1. Pass the already-fetched HTML to scripts/check_anchors.py through standard input.
  2. The script uses its own single-pass ContextRetentionParser; it does not fetch the URL or build a second DOM tree.
  3. Check for a substantive paragraph before the first <h2> and for fragment targets or in-page fragment links.
  4. Emit the script output unchanged.

Deterministic Grading Rubric

You must assign severities and generate findings strictly according to this matrix based on the JSON output from check_anchors.py.

Condition from Script Output Finding ID Finding Title Severity Priority
No 15-120 word paragraph appears before the first <h2> CTX-001 Missing Bottom-Line-First Paragraph medium medium
No in-page fragment links (href="#...") or section targets (id attributes on headings/sections) detected CTX-002 Missing Fragment Navigation for Deep Linking medium medium

Allowed Tools

  • bash (executing bundled script scripts/check_anchors.py)

Output

Emit only a valid JSON array conforming to the references/finding_schema.json format.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 1b012a3. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 2 months ago

README badge

README badge for tejasshukla2007/brand-ai-readiness-audit/context-retention-audit