Install
$ agentstack add skill-8dionysus-aoa-skills-aoa-change-protocol ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
aoa-change-protocol
Intent
Ground non-trivial changes in the owner route before editing, then keep the diff reviewable and explicitly verified without hidden scope expansion.
Trigger boundary
Use this skill when:
- the change affects code, config, docs, or operational surfaces in a meaningful way
- the task benefits from an explicit plan and verification path
- the task touches more than a trivial wording fix
- the request references previous work, sibling repositories, root surfaces, mechanics, generated outputs, or owner boundaries that must be inspected before planning
Do not use this skill when:
- the edit is tiny and has no meaningful review or operational consequence
- a more specific risk skill should be used instead
Inputs
- target goal
- owner repository or route surface
- touched files or surfaces
- source-of-truth or route-law surfaces that constrain the change
- optional prior-session evidence when the request asks how this repository,
tool, MCP, hook, validator, failure, or change pattern behaved before, preferably as a compact usage-chain packet when the question is about usage, outcomes, consequences, recurrence, or nearby errors for a stable anchor
- main risk
- intended validation path
- rollback idea
Outputs
- explicit plan
- scoped change
- named verification result
- post-change route review when source-of-truth, generated, decision, roadmap, changelog, quest, mechanic, or owner surfaces may have moved
- concise final report
Procedure
- choose the owner repository and read the nearest route card,
AGENTS.md, source-of-truth surfaces, and local validation path before planning - state the goal, touched surfaces, and evidence already inspected; if the request depends on prior work, inspect the current repo state rather than relying on memory
- when prior-session behavior matters for risk, speed, or recurrence, use
aoa-session-memory-evidence-route or an equivalent read-only aoa-session-memory-mcp packet; for a stable skill/MCP/hook/tool/script/test anchor, ask for usage-chain first when the question is how it was used, what happened after, or which failures/consequences followed; use entity-dossier when source identity, graph/cooccurrence/timeline context, related entities, or a heavier human packet is needed; expand to usage audit, neighborhood, literal, graph, goal, or answer routes only when the first packet is stale, truncated, missing refs, or too coarse; use the result as route evidence only
- identify the main risk, owner boundary, and rollback or recovery shape before editing
- prepare the smallest reviewable change that follows the current source route
- avoid unrelated cleanup, opportunistic refactors, or importing a sibling owner's authority into the local surface
- apply the change inside the declared scope
- run or name explicit verification, including generated/export rebuilds when source surfaces feed derived outputs
- perform the narrow post-change route review only for surfaces whose meaning actually moved
- report what changed, what was verified, what was skipped, what remains risky, and where the next owner route is
Contracts
- the change must remain reviewable
- verification must be explicit, not implied
- the change should stay inside the declared scope
- rollback thinking should exist before apply
- the plan should be grounded in inspected source surfaces, not only conversation memory or generic pattern recall
- generated, exported, compact, or derived surfaces should be refreshed from source rather than hand-authored as truth
- prior-session evidence can shape risk and route choice, but it does not
override current owner files, validators, or live runtime state
- cross-repo or mechanics changes should name owner boundaries and stop-lines before moving content
Risks and anti-patterns
- over-formalizing trivial edits
- symbolic verification that creates false safety
- using the report as a substitute for a readable diff
- silently expanding the task during implementation
- planning from stale memory when the repository already has route cards, mechanics, or source-of-truth docs that answer the question
- treating legacy, provenance, generated, or sibling-owner surfaces as if they were the active local contract
- inserting broad law blocks into local surfaces instead of shaping the change through the existing route
Verification
- confirm the owner route and relevant source surfaces were inspected before apply
- confirm the change stayed scoped
- confirm at least one explicit verification step was run or intentionally skipped with explanation
- if prior-session evidence was used, confirm the usage-chain, dossier, or
fallback route packet refs were named and current source/runtime evidence still controlled the change
- confirm generated or export surfaces were rebuilt when canonical inputs changed
- confirm the post-change route review touched only surfaces whose meaning moved
- confirm the report includes outcome, rollback thinking, skipped checks, and remaining owner-route risk
Technique traceability
Manifest-backed techniques:
- AOA-T-0001 from
8Dionysus/aoa-techniquesatcd276f040d55d490bd015b8698c7a5d594b9f875using pathtechniques/execution/agent-workflows-core/plan-diff-apply-verify-report/TECHNIQUE.mdand sections: Intent, When to use, Inputs, Outputs, Core procedure, Contracts, Risks, Validation - AOA-T-0002 from
8Dionysus/aoa-techniquesatcd276f040d55d490bd015b8698c7a5d594b9f875using pathtechniques/instruction/docs-boundary/source-of-truth-layout/TECHNIQUE.mdand sections: Intent, When to use, Inputs, Outputs, Core procedure, Contracts, Risks, Validation
Adaptation points
Project overlays should add:
- local source-of-truth files
- local validation commands
- local risk tiers
- approval or review rules
- local post-change route review surfaces such as changelog, roadmap, decision records, generated indexes, questbooks, or mechanics ledgers
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: 8Dionysus
- Source: 8Dionysus/aoa-skills
- License: Apache-2.0
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.