@research-engineering/agentic-proofkit 0.9.0 → 0.10.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/ADOPTION.md CHANGED
@@ -196,6 +196,67 @@ the proof binding already contains safe witness command facts and the caller
196
196
  provides witness command vocabulary. The projection avoids duplicated command
197
197
  identity; native execution and command freshness remain caller-owned.
198
198
 
199
+ ## Portable Agent Bootstrap
200
+
201
+ `REQ-PROOFKIT-WORKFLOW-016` through `REQ-PROOFKIT-WORKFLOW-018` own the bounded
202
+ Phase5A generation/check contract. Use `integration source --tool codex` or
203
+ `integration source --tool claude` through a repository-approved, already
204
+ installed launcher. JSON is the default and includes the exact generated
205
+ content, descriptor-owned target path, materialization identity, content and
206
+ consumed-capability digests, and separate metadata/body byte counts. Text is
207
+ the exact file content. The limits are 512 metadata bytes and 4096 body bytes,
208
+ not tokenizer-specific token counts.
209
+
210
+ The fixed repository paths are `.agents/skills/agentic-proofkit/SKILL.md` for
211
+ `codex` and `.claude/skills/agentic-proofkit/SKILL.md` for `claude`. Neither tool
212
+ is selected implicitly; selecting one does not inspect the other location.
213
+ The generator reads no repository files, writes nothing, and has no `--output`
214
+ or install option. Generated instructions contain no hooks or permission grants
215
+ and delegate policy, planning, and evidence semantics to current owners.
216
+ Resolve the approved installed launcher for each session. An absent or
217
+ ambiguous binding needs an owner decision, not an install, network fallback,
218
+ package-manager default, or persisted machine-local executable path.
219
+
220
+ For one manual export, first inspect the explicit repository root and every
221
+ component of the selected path. Stop on symlinks, aliases, unknown ownership,
222
+ local edits, or an existing target; do not overwrite it. The example below
223
+ assumes the owner has approved the npm launcher, inspected and prepared the
224
+ ordinary parent directories, and ensured no concurrent writer can replace
225
+ them. Replace the example absolute root with that inspected root. The shell,
226
+ not Proofkit, creates the file; noclobber protects an existing final file but
227
+ does not establish race-free parent traversal or transactional installation.
228
+
229
+ ```bash
230
+ (set -o noclobber; npm exec --offline -- agentic-proofkit integration source --tool codex --format text > /absolute/inspected/repository/.agents/skills/agentic-proofkit/SKILL.md)
231
+ npm exec --offline -- agentic-proofkit integration check --tool codex --repo-root /absolute/inspected/repository
232
+ ```
233
+
234
+ Check a failed export before treating its output as usable: shell redirection
235
+ may leave an empty or partial new file if generation or transport fails.
236
+ Resolve such a file manually under repository ownership; check never repairs it.
237
+ The checker admits flags before I/O and reads only the selected fixed path
238
+ through an application-write-free confined lease with bounded reobservation.
239
+ Exit 0 and `current` mean exact generated-byte equality; exit 2 reports
240
+ `missing`, `stale`, or `invalid`; exit 1 reports an invocation or operational
241
+ error, including denied reads, ambiguous paths, observed changes, cancellation,
242
+ or cleanup failure. It never prints observed bytes, their digest, or caller
243
+ root paths. Source exits 0 on successful generation and 1 on error.
244
+
245
+ Generated identity binds the shared template, selected descriptor, and exact
246
+ consumed registered public invocation contracts. Package version is absent
247
+ from the generated bytes; version-only or unrelated-command changes preserve
248
+ identity when consumed projections remain unchanged. Shared native source
249
+ digests may conservatively invalidate freshness. This is materialization
250
+ freshness, not proof of every transitive runtime behavior.
251
+
252
+ Phase5B managed install/update/remove remains open. Manual export is not an
253
+ installer, rollback protocol, or proof of host activation. Installed npm/Python
254
+ integration proof requires actual carrier execution; source-only tests do not
255
+ discharge it. Native-host file discovery, body loading, and approved-launcher calls
256
+ are separate observations requiring isolated sessions and absent/altered-file
257
+ controls; a prompt directly requesting CLI execution is not skill-use evidence.
258
+ Removing a file does not revoke instructions already loaded into host context.
259
+
199
260
  ## Requirement, Contract, And Test Order
200
261
 
201
262
  The durable semantic source is the repository-owned requirement package:
package/NON_CLAIMS.md CHANGED
@@ -36,6 +36,24 @@ This repository state does not claim:
36
36
  Consuming repositories must keep their own specifications, proof bindings,
37
37
  native witnesses, CI gates, rollback policy, and rollout decisions.
38
38
 
39
+ Phase5A integration source/check implements only bounded portable generation
40
+ and read-only generated-byte freshness. It does not install, update, remove,
41
+ repair, activate host instructions, grant tool permissions, or authorize native
42
+ execution. Phase5B managed lifecycle remains open; manual shell export does not
43
+ close it. `current` neither proves semantic full proof nor guarantees stability
44
+ after return. Consumed registered-contract identity is not complete transitive
45
+ native-semantic identity, and byte budgets are not token counts.
46
+
47
+ Declared witness routes do not prove execution. Installed npm/Python integration
48
+ smokes and final frozen-tree closure require actual execution against the named
49
+ source and artifacts. Native-host discovery, body loading, and approved-launcher
50
+ invocation require separate isolated evidence; direct CLI calls cannot prove
51
+ skill use. Removing an on-disk file does not revoke already-loaded instructions.
52
+ Read-only checks do not suppress filesystem-owned access metadata updates or
53
+ exclude uncooperative writers. Cancellation after the final pre-emission check
54
+ cannot retract bytes accepted by an external stdout writer, and prefix-accepting
55
+ transport failure does not imply an atomic sink.
56
+
39
57
  Reference infrastructure boundary:
40
58
 
41
59
  ```text
@@ -41,6 +41,7 @@ owner boundaries. It is not a second command-family inventory.
41
41
 
42
42
  | Family | Main commands | Caller provides | Proofkit owns | Consumer owns | Output authority |
43
43
  |---|---|---|---|---|---|
44
+ | Agent integration generation/check | `integration source`, `integration check` | explicit `claude` or `codex` tool; explicit repository root for check only | one portable bounded bootstrap, exact consumed registered-contract identity, descriptor-owned path, and confined read-only generated-byte freshness | launcher admission, manual materialization, instruction ownership, host discovery/activation, permissions, native verification, and any future managed lifecycle | generated source or missing/current/stale/invalid freshness report; neither is installation or execution authority |
44
45
  | Project state navigation | `status`, `next` | explicit repository root | bounded transaction-first materialized-project inspection, normalized observation identity, deterministic project-state classification, and one non-executable next action; admitted in-bound records bind exact content digests, while unread out-of-bound records intentionally identify only their invalid class | repository policy, byte identity for unread out-of-bound records, witness execution, receipt trust/currentness/scope, merge, release, deployment, rollout, and production readiness | project-status report, next-action packet, or bounded text projection |
45
46
  | Agent workflow planning | `change plan`, `native-evidence-guidance` | explicit checkpoint, completed stage ids, bounded context refs, governing authority ref, and required context ref ids | optional built-in `proofkit.reviewed-change.v1` checkpoint relation, reference-closed next-stage context, deterministic agent prompts, bounded text/JSON/envelope projections, and repository-neutral native-evidence guidance with closed applicability classes | custom workflow topology, repository state discovery, stage execution, native witness semantics, evidence collection, review conclusions, merge, release, deployment, and rollout authority | next-action plan, terminal workflow report, bounded agent envelope, or guidance catalog |
46
47
  | Adoption and scaffolding | `adopt plan`, `adopt materialize plan`, `adopt materialize apply`, `adopt materialize recover`, `repository-inventory`, `adoption-contract-envelope`, `adoption-workflow-plan`, `adoption-checklist`, `adoption-doctor`, `gradual-adoption`, `gradual-adoption-bootstrap`, `gradual-adoption-guidance`, `capability-map-admission`, `pilot-admission`, `scaffold-profile-plan`, `scaffold-project-structure`, `stack-preset` | explicit repository root, explicit fresh/code-baseline/audit-from-code intent, optional stack hint, owner-reviewed candidate packet, expected transaction and desired-state identities, recovery action, aggregate adoption contract envelope, checklist facts, target paths, owner routes, caller-extracted stale authority vocabulary facts, explicit pre-spec capability observations, and pilot records | bounded fixed-catalog root inventory, candidate-only front-door tasks, owner-closed read-only materialization plans, confined transactional apply and recovery receipts, aggregate contract-envelope admission, deterministic starter plans, checklist/report admission, bounded guidance envelopes, dry-run manifests, pre-spec trust-mode admission, adoption gap and stale-authority classification, and pilot shape admission | stack selection, arbitrary source inspection, candidate review, final requirement meaning, proof adequacy, rollout policy, text extraction, code observation extraction, and pilot truth | inventory, candidate-only plan, transaction-bound materialization plan or receipt, selected child output, report, seed packet, or agent envelope |
@@ -131,6 +132,7 @@ Semantic context routes are `requirement-context-compose`,
131
132
 
132
133
  | State or goal | Next Proofkit route | Stop or escalation condition |
133
134
  |---|---|---|
135
+ | A repository owner requests a portable Proofkit bootstrap or its exact-byte freshness. | `integration source --tool <tool>` generates JSON or exact file bytes with `--format text`; `integration check --tool <tool> --repo-root <caller-selected-root>` checks only the selected fixed path. Select `claude` or `codex` explicitly. | Source writes nothing; neither route accepts `--output` or install. Source exits 0/1 for success/error; check exits 0 for current, 2 for missing/stale/invalid, and 1 for invocation or operation error. Current proves only generated-byte equality. Use `ADOPTION.md` for inspected manual export; Phase5B installation/update/removal and host activation remain open. |
134
136
  | The repository may already contain a materialized Proofkit project. | `status --repo-root <caller-selected-root>` for the full bounded classification or `next --repo-root <caller-selected-root>` for its single action projection. | Treat `verification_required` as a request to run repository-owned verification, never as completion or approval. Stop on blocked or recovery-required states; the packet does not execute its route or own policy. |
135
137
  | The agent does not know where to start. | `adopt plan --mode <mode> --repo-root <caller-selected-root>`, where mode is `fresh`, `code-baseline`, or `audit-from-code` | Choose the trust intent explicitly. Treat the fixed-catalog inventory and tasks as a read-only candidate plan; stop before arbitrary source inspection, writing files, or making requirements authoritative. |
136
138
  | An agent needs a bounded, deterministic stage transition for an engineering change. | `change plan` selects the optional built-in `proofkit.reviewed-change.v1` profile; use `--agent-envelope` for the compact work packet and `native-evidence-guidance` when the consuming repository has not yet materialized repository-specific evidence instructions. | Supply only explicit current checkpoint, completed stages, and admitted context references. Apply conditional guidance slots only when their applicability class matches a declared consumer mechanism. Stop before treating the profile, plan, or guidance as repository policy or as proof that a stage ran, evidence exists, review passed, or merge/release is authorized. |
@@ -24,6 +24,11 @@ The public capability is deliberately small:
24
24
  bounded action. Admitted in-bound records bind exact content digests;
25
25
  unread out-of-bound records bind only their invalid class. Neither command
26
26
  claims native execution or proof completion.
27
+ 5. `integration source --tool <claude|codex>` emits one portable static
28
+ bootstrap without reading repositories or writing files;
29
+ `integration check --tool <claude|codex> --repo-root <path>` compares only
30
+ the selected descriptor-owned file through confined read-only inspection.
31
+ These are generation/check targets, not managed integration support.
27
32
 
28
33
  The change planner and evidence-guidance cores are stateless pure projections.
29
34
  Project status reads only an explicit repository root, the conventional routing
@@ -92,11 +97,47 @@ projections remain independently owned by the spec-proof-core package.
92
97
  without claiming cancellation rollback or atomicity from an external sink,
93
98
  and a versioned breaking replacement
94
99
  of the flat change route by `change plan` across source and installed carriers.
100
+ - `REQ-PROOFKIT-WORKFLOW-016`: one private descriptor/template owner, portable
101
+ approved-launcher guidance, 512-byte metadata and 4096-byte body limits, and
102
+ identity over exactly consumed registered invocation contracts rather than
103
+ package version or the entire command universe.
104
+ - `REQ-PROOFKIT-WORKFLOW-017`: pre-I/O invocation admission, a confined
105
+ application-write-free two-observation check with an 8192-byte read bound per
106
+ observation, distinct missing/current/stale/invalid states, non-disclosing
107
+ operational failures, and independent cancellation/cleanup outcomes.
108
+ - `REQ-PROOFKIT-WORKFLOW-018`: exact source/check CLI flags, JSON root shapes,
109
+ exit semantics, one source/check-only command family, and source, generated,
110
+ and installed-carrier closure obligations. A binding is not an execution
111
+ receipt; final frozen-tree and installed-carrier integration proof requires
112
+ actual execution against the named source and artifacts.
95
113
 
96
114
  Shared stable-JSON/diagnostic hardening is owned by the supply-chain-quality
97
115
  spec. Typed local-reference closure is owned by the existing agent-envelope
98
116
  requirement. This workflow spec consumes those owners without duplicating them.
99
117
 
118
+ ## Thin Integration Boundary
119
+
120
+ The production owner is `internal/command/agentintegration`. The descriptor
121
+ alone selects `.agents/skills/agentic-proofkit/SKILL.md` for `codex` or
122
+ `.claude/skills/agentic-proofkit/SKILL.md` for `claude`. Source and check share
123
+ the same renderer. Common name/description frontmatter and static Markdown
124
+ delegate to the current CLI instead of embedding schemas or workflow policy.
125
+ The bootstrap persists logical routes, not machine-local executable paths;
126
+ an absent or ambiguous repository-approved installed launcher needs an owner
127
+ decision, never an implicit package manager, install, or network fallback.
128
+
129
+ The source contract's default JSON includes exact content and its digest;
130
+ `--format text` emits those file bytes. Check returns only expected generated
131
+ identity and a bounded freshness classification, never observed content, its
132
+ digest, or caller root paths. Exit 0 means current, exit 2 means a classified
133
+ missing/stale/invalid file, and exit 1 means an invocation or operation error.
134
+ Source returns 0 on success and 1 on error. Neither accepts `--output` or an
135
+ install option. Manual export guidance is owned by `ADOPTION.md`.
136
+
137
+ Phase5B managed installation, update, and safe removal remains explicitly open.
138
+ Host discovery, body loading, and approved-launcher invocation require separate
139
+ isolated native-host observations; direct CLI execution cannot prove skill use.
140
+
100
141
  ## Non-Claims
101
142
 
102
143
  - Generated prompts are not authenticated and need not be obeyed or adequate.
@@ -106,5 +147,9 @@ requirement. This workflow spec consumes those owners without duplicating them.
106
147
  - Least dependency closure does not prove context truth, freshness, semantic
107
148
  sufficiency, or global minimality.
108
149
  - A finite pilot does not prove universal repository fit.
150
+ - Bootstrap byte bounds are not token counts, and consumed registered-contract
151
+ identity is not complete transitive semantic proof.
152
+ - Generated-byte freshness does not install, activate, authorize, or grant
153
+ permissions. File removal does not revoke instructions already in host context.
109
154
  - This spec does not select a persisted requirement-source codec.
110
155
  - This spec does not approve merge, release, rollout, or production readiness.
@@ -203,6 +203,57 @@
203
203
  "lifecycle": {"state": "active", "replacementRequirementIds": [], "evidenceRefs": []},
204
204
  "deferral": null,
205
205
  "updatePolicy": {"reviewOwnerId": "proofkit.agent-workflow", "requiresImpactDeclaration": true, "requiresProofBindingReview": true}
206
+ },
207
+ {
208
+ "requirementId": "REQ-PROOFKIT-WORKFLOW-016",
209
+ "ownerId": "proofkit.agent-workflow",
210
+ "invariant": "One private production owner in internal/command/agentintegration holds the finite tool/path descriptor table and one shared static bootstrap renderer consumed by both source and check; explicit codex selects .agents/skills/agentic-proofkit/SKILL.md and explicit claude selects .claude/skills/agentic-proofkit/SKILL.md, without a default tool or access to the other location. Generation is a deterministic repository-read-free and application-write-free projection with only name/description frontmatter, at most 512 UTF-8 metadata bytes and 4096 UTF-8 body bytes, exact file bytes in text mode, and no hooks, shell injection, permission grants, persona, background process, package installation, copied schema, state/action table, or independent workflow policy. The shared body delegates status, next, adoption/change planning, brief agent-route and existing detail retrieval, help, and native-evidence-guidance to current CLI owners. Identity binds the template, selected descriptor, and exact consumed registered routes and invocation contracts from existing app descriptors and generated input/output contract metadata; package version is absent from generated bytes, and version-only or unconsumed-command changes preserve bytes and identity when consumed registered projections are unchanged. Consumed contract or route changes invalidate identity; shared native source digests may conservatively invalidate freshness. Persist logical routes only, resolve the repository-approved already-installed launcher at invocation time, and stop for an owner decision when its binding is absent or ambiguous; tool, distribution channel, launcher profile, and concrete interpreter remain distinct coordinates without a persistent carrier option or network fallback.",
211
+ "claimLevel": "blocking",
212
+ "riskClass": "high",
213
+ "proofBindingRefs": ["proofkit/requirement-bindings.json"],
214
+ "nonClaimRefs": ["NC-PROOFKIT-WORKFLOW-016"],
215
+ "nonClaims": [
216
+ "Byte budgets do not prove tokenizer-specific token cost or model behavior.",
217
+ "Generation does not install instructions, activate a host skill, grant permissions, authorize execution, or prove native verification.",
218
+ "Materialization identity binds consumed registered invocation contracts, not complete schemas or every transitive native semantic behavior."
219
+ ],
220
+ "lifecycle": {"state": "active", "replacementRequirementIds": [], "evidenceRefs": []},
221
+ "deferral": null,
222
+ "updatePolicy": {"reviewOwnerId": "proofkit.agent-workflow", "requiresImpactDeclaration": true, "requiresProofBindingReview": true}
223
+ },
224
+ {
225
+ "requirementId": "REQ-PROOFKIT-WORKFLOW-017",
226
+ "ownerId": "proofkit.agent-workflow",
227
+ "invariant": "Integration freshness admits explicit tool, root, format, and flag cardinality before filesystem I/O, obtains expected bytes and the fixed selected path only from the integration source owner, and performs no application writes to selected, other-tool, adjacent, or global instructions. Reuse one read-only repository inspection lease and exact component-wise non-symlink regular-file traversal; check file kind before any content read and never read FIFO/device contents. Each of two complete observations reads at most 8192 bytes; reobserve the complete bounded state and bytes, verify file/route and pinned-root identity, and fail operationally on observed change. Exact entry or ordinary parent absence is missing; bounded regular UTF-8 text equal to generated bytes is current; differing bounded regular UTF-8 text, including unknown or old content, is stale; non-regular entry, symlink component, oversized file, NUL, or invalid UTF-8 is invalid. Permission, I/O, cleanup, portable-alias, changed-root/route, or cancellation failure is an operation error, never a classified success. Cancellation and file/lease cleanup outcomes participate independently in every result and dominate classification. Public reports and diagnostics disclose neither observed content, its digest, nor caller root paths; fixed descriptor-owned relative target paths and expected generated identities are public. A marker or stored digest never substitutes for checking actual bounded bytes.",
228
+ "claimLevel": "blocking",
229
+ "riskClass": "high",
230
+ "proofBindingRefs": ["proofkit/requirement-bindings.json"],
231
+ "nonClaimRefs": ["NC-PROOFKIT-WORKFLOW-017"],
232
+ "nonClaims": [
233
+ "Application-write-free inspection does not prevent filesystem-owned read metadata updates, exclude an uncooperative same-user writer, or identify unread oversized-file bytes beyond the bounded observation.",
234
+ "Current means exact generated-byte freshness only, not installation, host activation, approved-launcher invocation, native execution, semantic proof completion, or stability after return."
235
+ ],
236
+ "lifecycle": {"state": "active", "replacementRequirementIds": [], "evidenceRefs": []},
237
+ "deferral": null,
238
+ "updatePolicy": {"reviewOwnerId": "proofkit.agent-workflow", "requiresImpactDeclaration": true, "requiresProofBindingReview": true}
239
+ },
240
+ {
241
+ "requirementId": "REQ-PROOFKIT-WORKFLOW-018",
242
+ "ownerId": "proofkit.agent-workflow",
243
+ "invariant": "Exactly two no-input commands expose Phase5A: integration-source through integration source with built_in_package_catalog scope, and integration-check through integration check with explicit_filesystem_scan scope. Both require exactly one --tool from claude/codex, admit at most one --format from json/text with JSON default, reject input transport, surplus operands, color, carrier, output-file, and install options; check additionally requires exactly one --repo-root, which source rejects. Source JSON has exactly schemaVersion, kind, tool, targetPath, integrationId, content, contentDigest, capabilityDigest, metadataBytes, bodyBytes, and nonClaims; check JSON has exactly schemaVersion, kind, tool, targetPath, integrationId, expectedContentDigest, state, and nonClaims. Both schema versions are 1; kinds are proofkit.integration-source.v1 and proofkit.integration-check.v1. Source text is the exact generated content. Source exits 0 on success and 1 on error; check exits 0 only for current, 2 for missing/stale/invalid classified reports with empty stderr, and 1 for operation or invocation error. Cancellation observed before emission and serialization/cleanup failures produce no success report. Descriptor, dispatch, help, the agent-integrations source/check-only family, root-shape definitions and their hashes, CLI contracts, contract-map routes, exact native witnesses, generated metadata, and installed npm/Python carriers must close over the same bounded invocation/output relation without a second semantic owner; native source bindings include internal/command/agentintegration. Source-carrier independence and installed-carrier execution are separate obligations. Phase5B managed install/update/remove remains open and is not discharged by generation, check, manual export, or this contract closure.",
244
+ "claimLevel": "blocking",
245
+ "riskClass": "high",
246
+ "proofBindingRefs": ["proofkit/requirement-bindings.json"],
247
+ "nonClaimRefs": ["NC-PROOFKIT-WORKFLOW-018"],
248
+ "nonClaims": [
249
+ "Cancellation after the final pre-emission checkpoint cannot retract accepted output, and a caller-provided writer that accepts a prefix before failure is not an atomic sink.",
250
+ "Installed CLI behavior does not prove native-host file discovery, body loading, approved-launcher calls, registry publication, consumer migration, or host instruction revocation after file removal.",
251
+ "Phase5A is generation/check only, not managed installation, update, removal, host activation, permission grant, semantic full proof, merge approval, or production readiness.",
252
+ "Witness selectors declare exact verification routes, not executed or passed proof; installed npm/Python integration witnesses and final frozen-tree closure require actual execution against the named source and artifacts."
253
+ ],
254
+ "lifecycle": {"state": "active", "replacementRequirementIds": [], "evidenceRefs": []},
255
+ "deferral": null,
256
+ "updatePolicy": {"reviewOwnerId": "proofkit.agent-workflow", "requiresImpactDeclaration": true, "requiresProofBindingReview": true}
206
257
  }
207
258
  ],
208
259
  "nonClaims": [
@@ -676,7 +676,7 @@
676
676
  {
677
677
  "requirementId": "REQ-PROOFKIT-SPEC-033",
678
678
  "ownerId": "proofkit.spec-proof-core",
679
- "invariant": "The repository-transaction owner confines every effect to one explicit repository root; freezes a bounded canonical execution plan before effects; rejects non-canonical, reserved, symlinked, case-folding, Unicode-folding, and prefix-alias target identities; keeps staged payloads, backups, ownership records, and publish temporaries in one private transaction namespace; records each created directory by exact filesystem identity; and performs each target replacement by same-filesystem atomic rename. A bounded durable journal identifies the exact before-state prefix, applied-target count, and terminal result; apply fails closed on stale state, unknown control records, and cooperative concurrency; rollback removes only transaction-owned artifacts and restores exact bytes and modes; resume and rollback are state-compatible, action-stable, cancellation-aware, and replay-idempotent; terminal receipts preserve the complete observable result until a later valid transaction replaces them. These guarantees cover process interruption at every injected mutation boundary but do not claim filesystem-wide atomic visibility, power-loss durability beyond successful synchronization, protection from arbitrary readers, or safety against a non-cooperative same-user process mutating the private namespace.",
679
+ "invariant": "The repository-transaction owner confines every effect to one explicit repository root; freezes a bounded canonical execution plan before effects; rejects non-canonical, reserved, symlinked, case-folding, Unicode-folding, and prefix-alias target identities; keeps staged payloads, backups, ownership records, and publish temporaries in one private transaction namespace; records each created directory by exact filesystem identity; and performs each target replacement by same-filesystem atomic rename. A bounded durable journal identifies the exact before-state prefix, applied-target count, and terminal result; apply fails closed on stale state, unknown control records, and cooperative concurrency; rollback removes only transaction-owned artifacts and restores exact bytes and modes; resume and rollback are state-compatible, action-stable, cancellation-aware, and replay-idempotent; terminal receipts preserve the complete observable result until a later valid transaction replaces them. Native construction binds Apply to the complete transaction identity; public JSON re-admission remains descriptive even for empty payloads, while recovery retains its separate confined journal and staged-object admission. These guarantees cover process interruption at every injected mutation boundary but do not claim filesystem-wide atomic visibility, power-loss durability beyond successful synchronization, protection from arbitrary readers, or safety against a non-cooperative same-user process mutating the private namespace.",
680
680
  "claimLevel": "blocking",
681
681
  "riskClass": "critical",
682
682
  "proofBindingRefs": ["proofkit/requirement-bindings.json"],
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "@research-engineering/agentic-proofkit",
3
3
  "description": "Reusable proof profile, report, graph, and witness-planning primitives.",
4
- "version": "0.9.0",
4
+ "version": "0.10.1",
5
5
  "type": "module",
6
6
  "license": "MIT",
7
7
  "sideEffects": false,