From f57a1383c8c11a722117fcddd45c7926ee74664f Mon Sep 17 00:00:00 2001 From: Joe Feser Date: Fri, 12 Jun 2026 17:13:32 -0500 Subject: [PATCH 1/3] Phase 896/897 HACP exact docs examples edit --- ...p-is-not-runtime-transport-self-running.md | 44 ++++++++++++++++ .../proof-is-evidence-not-approval.md | 29 +++++++++++ .../bounded-action-report.md | 45 +++++++++++++++++ .../docs-backed-product-readback.md | 38 ++++++++++++++ .../manual-approved-loop-v0/evidence-chain.md | 42 ++++++++++++++++ .../human-decision-gate.md | 50 +++++++++++++++++++ .../no-retry-no-autonomous-continuation.md | 35 +++++++++++++ 7 files changed, 283 insertions(+) create mode 100644 docs/concepts/manual-loop-is-not-runtime-transport-self-running.md create mode 100644 docs/concepts/proof-is-evidence-not-approval.md create mode 100644 docs/examples/manual-approved-loop-v0/bounded-action-report.md create mode 100644 docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md create mode 100644 docs/examples/manual-approved-loop-v0/evidence-chain.md create mode 100644 docs/examples/manual-approved-loop-v0/human-decision-gate.md create mode 100644 docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md diff --git a/docs/concepts/manual-loop-is-not-runtime-transport-self-running.md b/docs/concepts/manual-loop-is-not-runtime-transport-self-running.md new file mode 100644 index 0000000..b4c4a8b --- /dev/null +++ b/docs/concepts/manual-loop-is-not-runtime-transport-self-running.md @@ -0,0 +1,44 @@ +# Manual Loop Is Not Runtime, Transport, Or Self-Running Evidence + +Manual approved-loop evidence is not runtime, transport, self-running, +production, compliance, certification, product-completion, loop-completion, or +formal-standardization evidence. + +## What The Manual Loop Shows + +The manual approved-loop example can show that a human-approved task packet led +to bounded documentation work, that the resulting report returned to a human +decision gate, and that retry, second-attempt, autonomous-continuation, +transport/runtime, durable-persistence, hosted-shell, and model/tool-call counts +remained `0`. + +## What The Manual Loop Does Not Show + +The manual approved-loop example does not show: + +- runtime execution; +- transport dispatch; +- HTTP hooks; +- RabbitMQ; +- watcher or polling behavior; +- durable persistence; +- hosted shell execution for bounded action; +- model or tool calls for bounded action; +- retry; +- second attempt; +- autonomous continuation; +- production readiness; +- compliance readiness; +- certification readiness; +- product completion; +- loop completion; +- formal standardization. + +Any future work in those areas requires separate human approval and separate +evidence. + +## Related Files + +- [No retry, no autonomous continuation](../examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md) +- [Bounded action report](../examples/manual-approved-loop-v0/bounded-action-report.md) +- [Proof is evidence, not approval](./proof-is-evidence-not-approval.md) diff --git a/docs/concepts/proof-is-evidence-not-approval.md b/docs/concepts/proof-is-evidence-not-approval.md new file mode 100644 index 0000000..4f7db97 --- /dev/null +++ b/docs/concepts/proof-is-evidence-not-approval.md @@ -0,0 +1,29 @@ +# Proof Is Evidence, Not Approval + +Proof is evidence, not approval. + +Evidence can support a later human decision, but it does not itself authorize +broader scope, implementation, publication, runtime execution, transport +execution, or a new lane of work. + +## Approval Boundary + +- A proof artifact can show custody, integrity, or observed work. +- A report can describe what happened under approved scope. +- A readback can summarize evidence for a human reviewer. +- A human decision record approves consequential next steps. + +## Non-Claims + +Proof is not production readiness evidence, compliance evidence, certification +evidence, product-completion evidence, loop-completion evidence, or formal +standardization evidence. + +Manual approved-loop evidence is not runtime evidence, transport evidence, or +self-running evidence. + +## Related Files + +- [Manual approved-loop evidence chain](../examples/manual-approved-loop-v0/evidence-chain.md) +- [Human decision gate](../examples/manual-approved-loop-v0/human-decision-gate.md) +- [Manual loop is not runtime/transport/self-running](./manual-loop-is-not-runtime-transport-self-running.md) diff --git a/docs/examples/manual-approved-loop-v0/bounded-action-report.md b/docs/examples/manual-approved-loop-v0/bounded-action-report.md new file mode 100644 index 0000000..9823f81 --- /dev/null +++ b/docs/examples/manual-approved-loop-v0/bounded-action-report.md @@ -0,0 +1,45 @@ +# Bounded Action Report + +A bounded action report is evidence about what happened under approved scope. It +is not permission to continue, retry, publish, standardize, or execute a new +lane. + +## Example Report + +```yaml +bounded_action_report: + report_id: bar_manual_loop_v0_example + source_task_packet: task_packet_manual_loop_v0_example + approved_scope: + - "Create documentation-only evidence examples." + - "Keep copy vendor-neutral and non-hype." + - "Preserve proof-is-evidence-not-approval wording." + completed_under_scope: + - "Recorded the manual approved-loop evidence chain." + - "Recorded count boundaries for retries, continuations, and runtime work." + - "Returned evidence to the human decision gate." + not_performed: + - runtime execution + - transport execution + - HTTP hooks + - RabbitMQ + - watcher or polling behavior + - durable persistence + - hosted shell execution + - model or tool calls + - retry + - second attempt + - autonomous continuation + requested_next_state: human_decision_required +``` + +## Boundary + +The report can support later review. It cannot approve another action, broaden +authority, or prove production/compliance/certification readiness. + +## Related Files + +- [Human decision gate](./human-decision-gate.md) +- [No retry, no autonomous continuation](./no-retry-no-autonomous-continuation.md) +- [Manual loop is not runtime/transport/self-running](../../concepts/manual-loop-is-not-runtime-transport-self-running.md) diff --git a/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md b/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md new file mode 100644 index 0000000..d5d519a --- /dev/null +++ b/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md @@ -0,0 +1,38 @@ +# Docs-Backed Product Readback + +A docs-backed product readback summarizes where the chain started, what was +approved, what was materialized, what returned to the human, which counts stayed +zero, and what was explicitly not authorized. + +## Readback + +The manual approved-loop v0 chain started with a human-approved task packet. One +local manual attempt produced one bounded action report under the approved +scope. The report returned evidence to a human decision gate. A later follow-up +action, separately approved by a human, produced one later follow-up action +report and stopped again at a human decision gate. + +The readback does not claim completion of the product, loop, runtime, transport, +self-running behavior, production readiness, compliance, certification, or +formal standardization. + +## Preserved Counts + +| Counter | Value | +| --- | ---: | +| First local manual attempt report count | 1 | +| Later follow-up action report count | 1 | +| Retry count | 0 | +| Second attempt count | 0 | +| Autonomous continuation count | 0 | +| Transport/runtime count | 0 | +| Durable persistence count | 0 | +| Hosted shell count for bounded action | 0 | +| Model/tool-call count for bounded action | 0 | + +## Related Files + +- [Evidence chain](./evidence-chain.md) +- [Human decision gate](./human-decision-gate.md) +- [No retry, no autonomous continuation](./no-retry-no-autonomous-continuation.md) +- [Proof is evidence, not approval](../../concepts/proof-is-evidence-not-approval.md) diff --git a/docs/examples/manual-approved-loop-v0/evidence-chain.md b/docs/examples/manual-approved-loop-v0/evidence-chain.md new file mode 100644 index 0000000..c8666e9 --- /dev/null +++ b/docs/examples/manual-approved-loop-v0/evidence-chain.md @@ -0,0 +1,42 @@ +# Manual Approved-Loop v0 Evidence Chain + +This example shows a local, manual approved loop. It starts with a +human-approved task packet, records bounded action/report evidence produced +under that authority, and returns to a human decision gate before any next lane +is approved. + +This is documentation evidence only. It is not runtime evidence, transport +evidence, self-running evidence, production evidence, compliance evidence, +certification evidence, product-completion evidence, loop-completion evidence, +or formal-standardization evidence. + +## Chain + +1. A human approves a bounded task packet. +2. A single local manual attempt produces one bounded action report. +3. The report records evidence about what happened under the approved scope. +4. The evidence returns to a human decision gate. +5. A later follow-up action, separately approved by a human, produces one + later follow-up action report. +6. The loop stops again at a human decision gate. + +## Count Boundaries + +| Counter | Value | +| --- | ---: | +| First local manual attempt report count | 1 | +| Later follow-up action report count | 1 | +| Retry count | 0 | +| Second attempt count | 0 | +| Autonomous continuation count | 0 | +| Transport/runtime count | 0 | +| Durable persistence count | 0 | +| Hosted shell count for bounded action | 0 | +| Model/tool-call count for bounded action | 0 | + +## Related Files + +- [Human decision gate](./human-decision-gate.md) +- [Bounded action report](./bounded-action-report.md) +- [No retry, no autonomous continuation](./no-retry-no-autonomous-continuation.md) +- [Proof is evidence, not approval](../../concepts/proof-is-evidence-not-approval.md) diff --git a/docs/examples/manual-approved-loop-v0/human-decision-gate.md b/docs/examples/manual-approved-loop-v0/human-decision-gate.md new file mode 100644 index 0000000..fbdb368 --- /dev/null +++ b/docs/examples/manual-approved-loop-v0/human-decision-gate.md @@ -0,0 +1,50 @@ +# Human Decision Gate + +A human decision gate records what is approved, what remains denied, what +evidence supports the decision, and which future work still requires a separate +approval. + +Evidence can inform the decision. Evidence does not approve the next step by +itself. + +## Decision Record Shape + +```yaml +decision_gate: + decision_id: hdg_manual_loop_v0_example + selected_decision: approve_bounded_docs_followup + approved_next_step: + description: "Apply the approved documentation-only follow-up." + scope: + - "Use the reviewed evidence chain." + - "Edit only the approved documentation paths." + denied_authority: + - runtime execution + - transport dispatch + - retry + - second attempt + - autonomous continuation + - production readiness claim + - compliance claim + - certification claim + - formal standardization claim + evidence_refs: + - bounded_action_report_manual_loop_v0 + - docs_backed_product_readback_manual_loop_v0 + requires_separate_future_approval: + - any runtime, transport, or self-running implementation + - any schema or protocol semantics change + - any production, compliance, certification, or standardization claim +``` + +## Boundary + +A human decision record approves consequential next steps. A report, proof, +receipt, checksum, or readback does not. + +## Related Files + +- [Evidence chain](./evidence-chain.md) +- [Bounded action report](./bounded-action-report.md) +- [Docs-backed product readback](./docs-backed-product-readback.md) +- [Proof is evidence, not approval](../../concepts/proof-is-evidence-not-approval.md) diff --git a/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md b/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md new file mode 100644 index 0000000..425946d --- /dev/null +++ b/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md @@ -0,0 +1,35 @@ +# No Retry, No Autonomous Continuation + +The manual loop preserves retry count `0`, second attempt count `0`, and +autonomous continuation count `0` unless a later human decision explicitly +approves different authority. + +## Boundary Counters + +```yaml +manual_loop_boundary: + retry_count: 0 + second_attempt_count: 0 + autonomous_continuation_count: 0 + transport_runtime_count: 0 + durable_persistence_count: 0 + hosted_shell_count_for_bounded_action: 0 + model_tool_call_count_for_bounded_action: 0 + next_state: human_decision_required +``` + +## Stop Rule + +After the bounded action report is created, the loop stops. Continuing requires +a human decision record that names the next approved step and the denied +authority that remains denied. + +This example does not introduce runtime, transport, HTTP hooks, RabbitMQ, +watcher/polling behavior, durable persistence, hosted shell execution, +model/tool calls, retry, a second attempt, or autonomous continuation. + +## Related Files + +- [Evidence chain](./evidence-chain.md) +- [Human decision gate](./human-decision-gate.md) +- [Manual loop is not runtime/transport/self-running](../../concepts/manual-loop-is-not-runtime-transport-self-running.md) From 6dd1f71ab0714bd39dfbee2a96afb2df97199d1f Mon Sep 17 00:00:00 2001 From: Joe Feser Date: Fri, 12 Jun 2026 17:18:03 -0500 Subject: [PATCH 2/3] Align manual loop examples with HACP vocabulary --- .../examples/manual-approved-loop-v0/bounded-action-report.md | 2 +- .../manual-approved-loop-v0/docs-backed-product-readback.md | 2 ++ docs/examples/manual-approved-loop-v0/human-decision-gate.md | 4 ++-- .../no-retry-no-autonomous-continuation.md | 2 +- 4 files changed, 6 insertions(+), 4 deletions(-) diff --git a/docs/examples/manual-approved-loop-v0/bounded-action-report.md b/docs/examples/manual-approved-loop-v0/bounded-action-report.md index 9823f81..c6148dd 100644 --- a/docs/examples/manual-approved-loop-v0/bounded-action-report.md +++ b/docs/examples/manual-approved-loop-v0/bounded-action-report.md @@ -30,7 +30,7 @@ bounded_action_report: - retry - second attempt - autonomous continuation - requested_next_state: human_decision_required + requested_next_step: request_human_decision ``` ## Boundary diff --git a/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md b/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md index d5d519a..4ae1740 100644 --- a/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md +++ b/docs/examples/manual-approved-loop-v0/docs-backed-product-readback.md @@ -6,6 +6,8 @@ zero, and what was explicitly not authorized. ## Readback +Readback id: `docs_backed_product_readback_manual_loop_v0`. + The manual approved-loop v0 chain started with a human-approved task packet. One local manual attempt produced one bounded action report under the approved scope. The report returned evidence to a human decision gate. A later follow-up diff --git a/docs/examples/manual-approved-loop-v0/human-decision-gate.md b/docs/examples/manual-approved-loop-v0/human-decision-gate.md index fbdb368..507ae6b 100644 --- a/docs/examples/manual-approved-loop-v0/human-decision-gate.md +++ b/docs/examples/manual-approved-loop-v0/human-decision-gate.md @@ -12,7 +12,7 @@ itself. ```yaml decision_gate: decision_id: hdg_manual_loop_v0_example - selected_decision: approve_bounded_docs_followup + selected_decision: accept_follow_up approved_next_step: description: "Apply the approved documentation-only follow-up." scope: @@ -29,7 +29,7 @@ decision_gate: - certification claim - formal standardization claim evidence_refs: - - bounded_action_report_manual_loop_v0 + - bar_manual_loop_v0_example - docs_backed_product_readback_manual_loop_v0 requires_separate_future_approval: - any runtime, transport, or self-running implementation diff --git a/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md b/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md index 425946d..345570b 100644 --- a/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md +++ b/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md @@ -15,7 +15,7 @@ manual_loop_boundary: durable_persistence_count: 0 hosted_shell_count_for_bounded_action: 0 model_tool_call_count_for_bounded_action: 0 - next_state: human_decision_required + next_status: needs_human_decision ``` ## Stop Rule From e10613084e412c68169ac536b9d6db7d37ff24cd Mon Sep 17 00:00:00 2001 From: Joe Feser Date: Fri, 12 Jun 2026 17:19:05 -0500 Subject: [PATCH 3/3] Clarify manual loop YAML sketches are illustrative --- .../manual-approved-loop-v0/bounded-action-report.md | 6 +++++- .../examples/manual-approved-loop-v0/human-decision-gate.md | 6 +++++- .../no-retry-no-autonomous-continuation.md | 4 ++++ 3 files changed, 14 insertions(+), 2 deletions(-) diff --git a/docs/examples/manual-approved-loop-v0/bounded-action-report.md b/docs/examples/manual-approved-loop-v0/bounded-action-report.md index c6148dd..735b19e 100644 --- a/docs/examples/manual-approved-loop-v0/bounded-action-report.md +++ b/docs/examples/manual-approved-loop-v0/bounded-action-report.md @@ -4,7 +4,11 @@ A bounded action report is evidence about what happened under approved scope. It is not permission to continue, retry, publish, standardize, or execute a new lane. -## Example Report +## Illustrative Report Sketch + +The YAML below is illustrative documentation, not a schema-valid HACP artifact. +Do not copy it as a base record; use the published schemas for validated +artifacts. ```yaml bounded_action_report: diff --git a/docs/examples/manual-approved-loop-v0/human-decision-gate.md b/docs/examples/manual-approved-loop-v0/human-decision-gate.md index 507ae6b..933968a 100644 --- a/docs/examples/manual-approved-loop-v0/human-decision-gate.md +++ b/docs/examples/manual-approved-loop-v0/human-decision-gate.md @@ -7,7 +7,11 @@ approval. Evidence can inform the decision. Evidence does not approve the next step by itself. -## Decision Record Shape +## Illustrative Decision Gate Sketch + +The YAML below is illustrative documentation, not a schema-valid HACP artifact. +Do not copy it as a base record; use the published schemas for validated +artifacts. ```yaml decision_gate: diff --git a/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md b/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md index 345570b..1e281f1 100644 --- a/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md +++ b/docs/examples/manual-approved-loop-v0/no-retry-no-autonomous-continuation.md @@ -6,6 +6,10 @@ approves different authority. ## Boundary Counters +The YAML below is illustrative documentation, not a schema-valid HACP artifact. +Do not copy it as a base record; use the published schemas for validated +artifacts. + ```yaml manual_loop_boundary: retry_count: 0