agentleFS
Sign inStart free
putervisionputervision/WebCrypt/.agents/skills/video-ingest/SKILL.md

video-ingest

Skill24 starsChanged 3 months agoMITView on GitHub

Teaches the agent to process, ingest, analyze, and compare WebM, MP4, and GIF video recordings using vision-memory-mcp and state-memory-mcp.

11 other projects use a version of this, 6 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use agent-reasoning-mcp, behavior-mcp and 1 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • putervisionSkill

    agent-reasoning-mcp

    Teaches the agent to use the Strategic Agent Reasoning MCP server for BDI goals, utility scoring, risk evaluation, and replanning.

    Also in 5 projects that use this

  • putervisionSkill

    behavior-mcp

    Teaches the agent to use the Behavior MCP server for ~60Hz in-browser behavior trees, triggers, and recordings.

    Also in 5 projects that use this

  • +1 more from putervision

The file

We split this skill into the rules it gives an agent

6 rules. 2 are common, 0 are distinctive, and 4 are very specific to this project.

  • 2

    Common

    Most of the 9 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Common:

    Use video ingestion whenever you encounter:

    line 14
  2. Common:

    When diagnosing bugs or linking test runs to task nodes:

    line 36
  3. Very specific:

    Ingest Video: Call manage_video (action: "ingest") with file path and action timestamps.

    line 38
  4. Very specific:

    Extract Evidence Payload: Read the returned evidence_payload containing source_video_id, frame_range, and timestamps_ms.

    line 39
  5. Very specific:

    Build Evidence Pack: Call create_evidence_pack linking keyframe_state_ids with state-memory-mcp task or blocker node IDs.

    line 40
  6. Very specific:

    Compare Trajectories (On Failure): Call compare_states(video_a_id: "...", video_b_id: "...") to pinpoint exact frame divergence points.

    line 41

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Video Frame Digesting & Temporal Memory Skill (video-ingest)

This skill provides step-by-step guidance, best practices, and operational patterns for digesting WebM, MP4, and GIF video recordings using @putervision/vision-memory-mcp.

1. When to Use Video Ingestion

Use video ingestion whenever you encounter:

  • E2E Playwright / Cypress / Selenium Test Artifacts: Recorded .webm screenchunks or .mp4 test run videos.
  • Bug Reproduction Videos: User-uploaded screen recordings demonstrating UI glitches or crashes.
  • UI Walkthrough Recordings: Demonstrations of complex multi-step user workflows.
  • Visual Regression Diagnostics: Comparing a passing baseline video run against a failing test run.

2. Ingestion Strategies & Parameter Tuning

ScenarioRecommended ParametersWhy
Action Event Timestamps (Highest Precision)action_timestamps: [1.2, 3.5, 7.0]Samples keyframes at exact interaction timestamps (clicks, types, navigation events) from test runners or state-memory logs.
Dynamic UI / Animationsscene_threshold: 0.3, fps: 1Combines scene-change detection (gt(scene,0.3)) with 1 fps background sampling to capture major screen transitions without frame bloat.
High-Speed Test Runsfps: 2 or fps: 5Increases frame rate sampling for rapidly switching UI test steps.
Long Screen Recordingsfps: 0.5, scene_threshold: 0.4Lowers sampling rate to conserve storage while extracting unique keyframe states.

3. Mandatory Dual-MCP Evidence Workflow

When diagnosing bugs or linking test runs to task nodes:

  1. Ingest Video: Call manage_video (action: "ingest") with file path and action timestamps.
  2. Extract Evidence Payload: Read the returned evidence_payload containing source_video_id, frame_range, and timestamps_ms.
  3. Build Evidence Pack: Call create_evidence_pack linking keyframe_state_ids with state-memory-mcp task or blocker node IDs.
  4. Compare Trajectories (On Failure): Call compare_states(video_a_id: "...", video_b_id: "...") to pinpoint exact frame divergence points.

4. CLI Quick Reference

# Ingest WebM / MP4 video recording with custom category
vision-memory-mcp video ingest ./recording.webm --category e2e_test

# Ingest with explicit action timestamps
vision-memory-mcp video ingest ./bug.mp4 --category bug_repro --action-timestamps 1.2,3.5,8.0

# Inspect chronological timeline & keyframes for a video ID
vision-memory-mcp video inspect vid_d7fc21c0ad52f861

# List all ingested video memory records
vision-memory-mcp video list

5. Consolidated Video MCP Tools Reference

  • manage_video (action: "ingest"): Ingests video file/base64, extracts keyframes, dHash deduplicates, and generates CLIP vector embeddings.
  • manage_video (action: "timeline"): Fetches step-by-step keyframes, exact timestamps (timestamp_ms), OCR snippets, and grounded target handles.
  • manage_video (action: "search"): Searches video memory by description query, category, tags, or file path.
  • compare_states (video_a_id, video_b_id): Calculates similarity score between two recordings and pinpoints exact timestamp divergence.
  • create_evidence_pack: Produces an immutable, cryptographically hashable evidence pack payload linking keyframes to task graph nodes.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free