Aisp creator evolution aisp

AISP Creator & Evolution (AISP-backed Agent Skill)

<!-- generated_from_aisp: true --> <!-- source: aisp.aisop.json --> <!-- generator: tools/aisp_skill_md.py -->

This SKILL.md is a thin Agent Skills discovery bridge, not the source of truth. The executable source of truth is the same-folder aisp.aisop.json AISP program.

Deleting this file does not change the native AISP skill. A conforming AISP/AISOP runtime should load aisp.aisop.json, read user.content.aisp_contract, and run user.content.aisop.main exactly as declared.

How to use

1. Load aisp.aisop.json from this folder.
2. Read user.content.aisp_contract before following any workflow.
3. Follow user.content.instruction: STRICTLY OBEY aisp_contract; its non_negotiable rules are inviolable; then RUN aisop.main.
4. Load declared resources only when the AISP graph reaches the node that needs them.
5. Enforce every non-negotiable rule through the mechanism named by enforced_by.

Declared resources

• schemas/skill-design.schema.json (schema, read_only)
• schemas/research-request.schema.json (schema, read_only)
• schemas/research-result.schema.json (schema, read_only)
• schemas/evolution-evidence.schema.json (schema, read_only)
• references/binding-rules.md (reference, read_only)
• README_CN.md (documentation, read_only)
• .gitignore (packaging, read_only)
• aisop_specification/README.md (spec_index, read_only)
• aisop_specification/MANIFEST.sha256.json (manifest, read_only)
• aisop_specification/aisop-spec.md (spec, read_only)
• aisop_specification/aisop-spec_CN.md (spec, read_only)
• aisp_specification/README.md (spec_index, read_only)
• aisp_specification/MANIFEST.sha256.json (manifest, read_only)
• aisp_specification/AISP_Protocol.md (spec, read_only)
• aisp_specification/AISP_Protocol_cn.md (spec, read_only)
• aisp_specification/aisp.proto (spec, read_only)
• aisp_specification/standards/AISP_Standard.core.aisop.json (standard, read_only)
• aisp_specification/standards/AISP_Standard.ecosystem.aisop.json (standard, read_only)
• aisp_specification/standards/AISP_Standard.security.aisop.json (standard, read_only)
• aisp_protocol_schemas/README.md (schema_index, read_only)
• aisp_protocol_schemas/MANIFEST.sha256.json (manifest, read_only)
• aisp_protocol_schemas/aisp-contract-v1.schema.json (schema, read_only)
• aisp_protocol_schemas/runtime-trace-v1.schema.json (schema, read_only)
• aisp_protocol_schemas/tool-capabilities-v1.schema.json (schema, read_only)
• aisp_protocol_schemas/registry-manifest-v1.schema.json (schema, read_only)
• aisp_reference_tools/README.md (tool_index, read_only)
• aisp_reference_tools/MANIFEST.sha256.json (manifest, read_only)
• aisp_reference_tools/aisp_validate.py (script, execute_only)
• aisp_reference_tools/aisp_hash.py (script, execute_only)
• aisp_reference_tools/aisp_readme.py (script, execute_only)
• aisp_reference_tools/aisp_skill_md.py (script, execute_only)
• aisp_reference_tools/aisp_check_runtime_trace.py (script, execute_only)
• aisp_reference_tools/aisp_validate_agent_skill_bridge.py (script, execute_only)
• aisp_reference_tools/check_doc_sync.py (reference, read_only)
• aisp_reference_tools/check_markdown_links.py (reference, read_only)
• aisp_reference_tools/__init__.py (reference, read_only)
• templates/skill-template.md (template, read_only)
• scripts/validate_aisp_skill.py (script, execute_only)
• scripts/simulate_aisp_flow.py (script, execute_only)
• scripts/check_generated_candidate.py (script, execute_only)
• evals/evals.json (eval, read_only)
• evals/script-behavior.json (eval, read_only)
• evals/runtime-traces/hard-pass.json (eval, read_only)
• evals/runtime-traces/hard-attested-no-event.json (eval, read_only)
• evals/runtime-traces/bad-shape.json (eval, read_only)

Non-negotiable boundaries

• Follow RUN aisop.main; route by intent and never skip research, validation, or human review. (aisop.main)
• Required research MUST run via the research sub-graph, not ad-hoc browsing. (run_research.step2:sys.assert)
• Design MUST NOT start until mandatory research has completed. (assert_research_complete.step1:sys.assert)
• A delivered skill MUST be a conformant AISP skill (passes M1-M6 and safety checks); reject otherwise. (validate_candidate.step5:sys.assert)
• MUST obtain explicit human approval before delivering a candidate. (review_candidate.step2:sys.io.confirm)
• Never write outside the resolved candidate root; never publish or perform destructive actions without confirmation. (tools)

Runtime boundary

Agent Skills platforms can use this bridge to discover and inspect the package. Hard guarantees such as sys.assert, sys.io.confirm, tool gating, dispatch behavior, and path confinement require a conforming AISP/AISOP runtime. A generic non-AISOP agent can only follow the contract on a best-effort basis.

Passing SKILL.md generation or bridge validation proves only projection consistency and bridge shape. It does not prove external trust, safety, registry approval, or hard execution on a non-AISOP platform.

Align Axiom 0: Human Sovereignty and Wellbeing. AISP - AI Skill Protocol V1.0.0. www.aisp.dev