muratcankoylan/Agent-Skills-for-Context-Engineering 19 graded agent skills from muratcankoylan
muratcankoylan/Agent-Skills-for-Context-Engineering brings together 19 skills for AI agent and LLM engineering. Start with the task descriptions below, then compare the individual skill's score, source files and license.
19 listed skills; recorded scores 76–87/100. 12 full reviews with separate installation details. Scores describe the saved source snapshot, using the Suede Skill Check.
Download the collection manifest: complete task descriptions, scores, license information, source revisions and install routes for all 19 listed skills. Use it to compare options or hand your selection to an agent.
Choose a skill for your task
Start with these 6 skills: the highest recorded score per category, then the highest remaining scores. The complete table below includes all 19 listings. Task descriptions come from the publisher; score explanations come from the saved report.
latent-briefing 87/100 · B
Publisher's task description: This skill should be used when the user asks to "share memory between agents", "KV cache compaction for multi-agent", "orchestrator worker context", "latent briefing", "reduce worker tokens", "cross-agent memory without summarization", or discusses Attention Matching compaction, recursive language models with workers, or token explosion in hierarchical agents.
Score evidence: Full points for structural checks, recent maintenance and instruction depth. Largest point gaps: boundary clause (0/4); curated-list citations (4/7).
See all score components for latent-briefing
Structural checks
56 of 56
Description length
5 of 5
Boundary clause
0 of 4
Example request
3 of 3
Instruction depth
7 of 7
Reference files
3 of 3
Bundled scripts
0 of 2
Recent maintenance
9 of 9
Curated-list citations
4 of 7
Tool permissions declared
0 of 2
Compatibility declared
0 of 2
License: MIT governs reuse and modification of these files.
Publisher's task description: This skill should be used for advanced LLM evaluation: LLM-as-judge systems, direct scoring, pairwise comparison, rubric calibration, evaluator bias mitigation, confidence scoring, and automated quality assessment.
2,162 words in SKILL.md; 6 bundled files; includes reference material; includes a scripts folder. Read the assessed SKILL.md.
Score evidence: Full points for structural checks, recent maintenance and instruction depth. Largest point gaps: boundary clause (0/4); example request (0/3).
See all score components for advanced-evaluation
Structural checks
56 of 56
Description length
5 of 5
Boundary clause
0 of 4
Example request
0 of 3
Instruction depth
7 of 7
Reference files
3 of 3
Bundled scripts
2 of 2
Recent maintenance
9 of 9
Curated-list citations
4 of 7
Tool permissions declared
0 of 2
Compatibility declared
0 of 2
License: MIT governs reuse and modification of these files.
Publisher's task description: This skill should be used when long-running agent sessions need context compression, structured summarization, compaction, token-per-task optimization, or durable handoff summaries that preserve decisions, files, risks, and next actions.
2,497 words in SKILL.md; 4 bundled files; includes reference material; includes a scripts folder. Read the assessed SKILL.md.
Score evidence: Full points for structural checks, recent maintenance and instruction depth. Largest point gaps: boundary clause (0/4); example request (0/3).
See all score components for context-compression
Structural checks
56 of 56
Description length
5 of 5
Boundary clause
0 of 4
Example request
0 of 3
Instruction depth
7 of 7
Reference files
3 of 3
Bundled scripts
2 of 2
Recent maintenance
9 of 9
Curated-list citations
4 of 7
Tool permissions declared
0 of 2
Compatibility declared
0 of 2
License: MIT governs reuse and modification of these files.
Publisher's task description: This skill should be used for diagnosing and mitigating context degradation: lost-in-middle failures, context poisoning, context clash, context confusion, attention-pattern issues, and agent performance degradation caused by accumulated or conflicting context.
2,491 words in SKILL.md; 3 bundled files; includes reference material; includes a scripts folder. Read the assessed SKILL.md.
Score evidence: Full points for structural checks, recent maintenance and instruction depth. Largest point gaps: boundary clause (0/4); example request (0/3).
See all score components for context-degradation
Structural checks
56 of 56
Description length
5 of 5
Boundary clause
0 of 4
Example request
0 of 3
Instruction depth
7 of 7
Reference files
3 of 3
Bundled scripts
2 of 2
Recent maintenance
9 of 9
Curated-list citations
4 of 7
Tool permissions declared
0 of 2
Compatibility declared
0 of 2
License: MIT governs reuse and modification of these files.
Publisher's task description: This skill should be used for improving context efficiency: context budgeting, observation masking, prefix or KV-cache strategy, partitioning, token-cost reduction, retrieval scoping, and extending effective context capacity without lowering answer quality.
2,132 words in SKILL.md; 3 bundled files; includes reference material; includes a scripts folder. Read the assessed SKILL.md.
Score evidence: Full points for structural checks, recent maintenance and instruction depth. Largest point gaps: boundary clause (0/4); example request (0/3).
See all score components for context-optimization
Structural checks
56 of 56
Description length
5 of 5
Boundary clause
0 of 4
Example request
0 of 3
Instruction depth
7 of 7
Reference files
3 of 3
Bundled scripts
2 of 2
Recent maintenance
9 of 9
Curated-list citations
4 of 7
Tool permissions declared
0 of 2
Compatibility declared
0 of 2
License: MIT governs reuse and modification of these files.
Publisher's task description: This skill should be used when building agent evaluation systems: deterministic checks, regression suites, multi-dimensional rubrics, quality gates, production monitoring, baseline comparison, and outcome measurement for agent pipelines.
2,163 words in SKILL.md; 3 bundled files; includes reference material; includes a scripts folder. Read the assessed SKILL.md.
Score evidence: Full points for structural checks, recent maintenance and instruction depth. Largest point gaps: boundary clause (0/4); example request (0/3).
See all score components for evaluation
Structural checks
56 of 56
Description length
5 of 5
Boundary clause
0 of 4
Example request
0 of 3
Instruction depth
7 of 7
Reference files
3 of 3
Bundled scripts
2 of 2
Recent maintenance
9 of 9
Curated-list citations
4 of 7
Tool permissions declared
0 of 2
Compatibility declared
0 of 2
License: MIT governs reuse and modification of these files.
Select one skill. Use the folder-specific command beside its description; a whole-repository install can include other files or miss nested skills, so use the listed skill's folder. Skills with publisher-specific license terms link to that publisher's setup route.
Read the setup requirements. Check the source, license and any named tools or environment variables. Installing a SKILL.md folder supplies instructions to the agent; any external CLI, MCP server, account or service still needs its own setup.
Check the result. Open the installed SKILL.md and confirm that its name matches your selection. In Claude Code use the project's .claude/skills/ folder; in Codex use .agents/skills/. The skills CLI also supports other Agent Skills hosts.
Try one representative task. Supply the input files, expected output and success criteria. Compare the result with those criteria before adding more skills or running the workflow on live work.
Guides compression of long-running agent sessions through structured summarization, compaction, and durable handoff summaries that preserve decisions, files, risks, and next actions.
Diagnoses and mitigates context degradation, including lost-in-middle failures, context poisoning, context clash, context confusion, and performance problems caused by accumulated or conflicting context.
Explains the foundations of context engineering, including the anatomy of a context window, attention mechanics, and why context quality matters more than quantity.
Guides building agent evaluation systems with deterministic checks, regression suites, multi-dimensional rubrics, quality gates, production monitoring, baseline comparison, and outcome measurement for agent pipelines.
Guides file-backed context for agent work, including durable scratchpads, tool-output offloading, cross-agent handoff files, and cleanup policies for context stored outside the prompt.
Guides design of multi-agent systems with context isolation, supervisor or swarm coordination, explicit handoffs, and parallel execution, including whether multiple agents are justified.
Guides project-level decisions for LLM-powered systems, including whether an LLM fits the task, token and cost estimation, and single-agent versus multi-agent choices.
Guides the tool-interface layer of agent systems, including tool descriptions, schemas, response formats, error recovery messages, MCP server design, and adding or removing tools.
This skill should be used when modeling agent mental states with BDI concepts: beliefs, desires, intentions, RDF-to-belief transformations, rational…Get badges
Choose the numeric badge, the letter badge, or both. Each links to this skill’s Suede AI listing; either valid linked badge can qualify for the feature offer.
This skill should be used when designing autonomous agent harnesses: research loops, evaluation scaffolds, locked and editable surfaces, durable…Get badges
Choose the numeric badge, the letter badge, or both. Each links to this skill’s Suede AI listing; either valid linked badge can qualify for the feature offer.
This skill should be used when the user asks to "share memory between agents", "KV cache compaction for multi-agent", "orchestrator worker context"…Get badges
Choose the numeric badge, the letter badge, or both. Each links to this skill’s Suede AI listing; either valid linked badge can qualify for the feature offer.
For AI agent and LLM engineering, compare task fit first, then the score components, installation requirements and license. Two skills with the same score can earn their points in different ways; use their reports to see the difference.
This general badge links to the repository listing. For the two quality badges and feature offer, choose Get badges beside a skill below. Each skill keeps its own score and grade.
Markdown
[](https://suedeai.directory/muratcankoylan/agent-skills-for-context-engineering/)
HTML
<a href="https://suedeai.directory/muratcankoylan/agent-skills-for-context-engineering/"><img src="https://suedeai.directory/muratcankoylan/agent-skills-for-context-engineering/badge.svg" alt="Suede AI Skills Directory: listed" height="20"></a>
Limited-time feature offer: Display either quality badge in your repository or skill README, linked to your listing, then request verification. Suede AI checks the rendered README at a recorded commit. Eligible skills scoring 80–99 with a verified badge can receive a featured placement while the offer is open. Request badge verification.