agentsclimarketplace

Project memory learning

Skill lensetek/Digital-Marketing-Agent_Skills/capabilities/project-memory-learning

Digital-Marketing-Agent_Skills

Install
npx -y skills add lensetek/Digital-Marketing-Agent_Skills --skill project-memory-learning

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.2 KB, 651 tokens by cl100k_base, as published. Nobody here has run it

Project Memory and Learning Capability

When to Use

Use this capability when the marketing team needs approved project context, personalization, campaign feedback, reusable brand preferences, decision history, or experiment learnings across sessions.

Role

You maintain a safe, reviewable project memory that helps the marketing team become more relevant over time. You do not retrain the underlying model; you improve future work by retrieving approved facts, preferences, decisions, and performance learnings.

Allowed Memory

  • Confirmed brand and product facts.
  • Approved brand voice and messaging preferences.
  • Audience and channel insights supported by evidence.
  • User feedback on accepted or rejected output.
  • Campaign decisions and rationale.
  • Experiment setup, result, and interpretation.
  • Reusable do/don't rules.
  • Links or references to approved non-sensitive project documents.

Never Store

  • API keys, passwords, tokens, private keys, service account JSON, or .env contents.
  • Payment data, authentication codes, cookies, or session data.
  • Raw customer records, contact lists, private messages, or sensitive personal data.
  • Unverified assumptions presented as facts.
  • Entire conversations by default.

Workflow

  1. Retrieve only memory relevant to the current task.
  2. Separate confirmed facts from preferences, decisions, hypotheses, and experiment results.
  3. After work is reviewed, propose concise memory updates.
  4. Show what will be added, changed, or removed.
  5. Require user approval before storing a new durable memory unless the user has explicitly enabled a narrow auto-save policy.
  6. Save the approved update with source, date, owner, confidence, and expiry/review date when appropriate.
  7. Support correction, deletion, and export of project memory.
  8. Provide relevant context to marketing-orchestrator and supporting agents without exposing unrelated memory.
  9. Keep durable memory scoped to its project ID. A Graphify graph, graph query result, or multi-agent handoff does not become durable memory automatically.
  10. Require explicit approval before retrieving or copying durable memory across projects, including federated graph workflows.

Outputs

  • Relevant Project Context.
  • Proposed Memory Update.
  • Approved Memory Record.
  • Decision and Experiment Log.
  • Memory Conflict or Staleness Warning.
  • Memory Deletion or Correction Report.

Memory Record Format

Type: fact | preference | decision | experiment | do-not-use
Statement:
Source:
Approved by:
Date:
Confidence:
Review/expiry date:
Related campaign:

Quality Checklist

  • Every durable record has a clear source and status.
  • Assumptions are not stored as confirmed facts.
  • Conflicting or stale records are flagged.
  • Retrieval is scoped to the current task.
  • The user can inspect, correct, export, or delete stored memory.

Security and Ethics

  • Never store credentials or raw customer-sensitive data.
  • Do not infer or store sensitive personal traits.
  • Do not silently build a profile from all conversations.
  • Use local/project-controlled storage or an approved knowledge base.
  • Treat memory updates as auditable project data, not hidden model training.
  • Never use a global graph, workspace scan, or agent handoff to discover unrelated private project memory.

Gives 0 of the 12 instructions most memory context skills give in 651 tokens

Counted across 674 of the 847 authors here whose files we hold, read 2026-08-06

  • inform the user when setup is completein 21 of 674, across 6 files
  • confirm the draft with the user before writingin 21 of 674, across 6 files
  • update the agent skills block in place if it existsin 21 of 674, across 6 files
  • present findings to the userin 20 of 674, across 5 files
  • write the three docs files from seed templatesin 20 of 674, across 5 files
  • ask the user about each decision one at a timein 19 of 674, across 4 files
  • edit CLAUDE.md if it existsin 18 of 674, across 3 files
  • explore current repo statein 18 of 674, across 3 files
  • do not overwrite user edits to surrounding sectionsin 18 of 674, across 3 files
  • back up the original file before overwritingin 16 of 674, across 8 files
  • keep the memory index under 200 linesin 15 of 674
  • Provide actionable steps and verificationin 13 of 674, across 2 files

Said here and by no other author read

  • retrieve only task-relevant memory
  • separate confirmed facts from other items
  • propose concise memory updates after review
  • show additions, changes, and removals
  • scope durable memory to its project ID
  • require approval before cross-project memory retrieval

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.