|
1 | 1 | type typeName facet quality_score quality_max quality_level title content |
2 | | -D Decision 5 5 strong Sessions Mirror Modes promoted to Tier-1 principle Each epistemic mode earns its own session because the structural blindness that makes a creator unable to validate their own work, a critic unable to remediate their own findings, and a planner unable to honestly execute their own plan is the same blindness in three different shapes. The principle generalizes critic-cannot-be-resolver (klappy://canon/constraints/critic-cannot-be-resolver) and verification-requires-fresh-context (klappy://canon/principles/verification-requires-fresh-context) to a universal claim: every gate between modes is a context boundary, and every context boundary deserves a fresh session. Five mode-bound roles named — explorer, planner, builder, validator, resolver — each mapping 1:1 to a canonical mode. |
3 | | -D Decision 5 5 strong Mode Transitions Require Encoded Handoff promoted to Tier-1 constraint Every transition between modes requires two artifacts: a journal entry recording the transition (universal, no exception) and a transition-specific minimal handoff (explorer→planner: synthesis ledger; planner→builder: plan with assumptions and scope; builder→validator: artifact + claims; validator→resolver: findings with dispositions; resolver→re-validator: revised artifact + remediation summary). Reversion and skip remain permitted with explicit acknowledgment per existing canon. New: operator override added as third permitted deviation — explicit runtime declaration of mode collapse under declared urgency, with stated reason, acknowledged risks, and journal entries at session start and end. |
4 | | -D Decision 5 5 strong Resolution promoted to fifth canonical epistemic mode The four-mode canon already implied a fifth phase: "fixes belong to iteration, which is a fresh execution pass scoped by validation findings." This patch makes the implied mode explicit. Resolution sits between Validation and a re-validation pass. Purpose: produce a revised artifact scoped strictly by validation findings. Truth condition: addresses findings without expanding scope. Primary risk: scope creep — using remediation as cover for redesign. Non-Collapse Rule extended with three resolution-specific invalid moves. Frontmatter date bumped to 2026-05-10; epoch to E0008.5; stability stays semi_stable since the addition is canonical implication-made-explicit, not a new claim about how truth is formed. |
5 | | -D Decision 5 5 strong Handoff-Insufficiency Signaling added as runtime feature #8 The runtime cannot guarantee handoff quality — that is upstream encoding-norm discipline. But the runtime can ensure that bad handoffs are detectable. A receiving session that cannot proceed on an insufficient handoff returns a structured "handoff_insufficient" outcome with named missing items and a proposed_resolution (request fresh upstream session / request clarification / operator override). The runtime records the insufficiency in the journal, blocks the downstream handoff, and surfaces the resolution path to the consumer. Audit trails accumulate; chronic insufficiency at a specific gate signals norm gaps that the project's encoding discipline must address. |
6 | | -L Learning 5 5 strong The architecture only pays back when handoffs preserve dynamic types Operator surface insight: "If encodings and other handoffs are lacking and of poor quality or missing key dynamic types any form of missing details and crucial context excluded will cause problems that may not outweigh the benefit of exclusive sessions." Handoff quality is the architecture's load-bearing assumption. Specific failure modes named: lossy encoding (decisions captured but not live tensions); missing dynamic types (static plan but not live state); missing crucial context that felt obvious to the encoder but wasn't written down because everyone-on-the-project-knew. When these dominate, session-per-mode becomes worse than mode-collapse: orchestration cost paid in full and receiving sessions operating on distorted views. Failure-mode sections added to sessions-mirror-modes principle (substantive), encoded-handoff constraint (pointer + summary), and persona-shaped-agent-runtime (handoff-insufficiency signaling feature). |
7 | | -C Constraint 5 5 strong Encoding norms must emerge from observed failures The constraint codifies handoff PRESENCE; QUALITY is upstream encoding-norms discipline that the constraint requires the existence of but cannot measure. Type-system / runtime-validation pattern: encoded-handoff constraint catches some classes of failure statically (handoff exists with minimum content); handoff-insufficiency signaling catches the rest at runtime (receiving session structurally cannot do its job). A project that observes its planner-role sessions routinely flagging explorer-handoffs as insufficient on a specific dimension has evidence that its synthesis-ledger encoding norms need extension. The runtime surfaces failure patterns; the project refines norms. |
8 | | -O Observation 5 5 strong Parallelism patterns named as four cases with one forbidden Within-mode parallelism (multiple explorers, multiple validators with different lenses, multiple builders on independent scope) — encouraged. Multi-participant single session (builder + assistant pair-programming) — allowed within a mode. Cross-mode on the same artifact — FORBIDDEN; the only pattern blocked. Cross-mode on different artifacts — fully independent, default mode of a productive multi-agent system. Distinguishing question: are these sessions operating on the same artifact in different modes simultaneously? If yes, forbidden; if no, allowed. Runtime enforces the forbidden pattern by refusing validator sessions on artifacts whose builder hasn't produced an encoded handoff. |
9 | | -H Handoff 5 5 strong PR #187 update — five new sections + new principle + new constraint + epistemic-modes Resolution patch klappy.dev branch claude/runtime-canon-acd1 / PR #187 picks up: (1) new canon/principles/sessions-mirror-modes.md (Tier-1 principle, 25kb), (2) new canon/constraints/mode-transitions-require-encoded-handoff.md (Tier-1 constraint, 21kb), (3) modified canon/definitions/epistemic-modes.md (Resolution as fifth mode, Non-Collapse Rule extended, frontmatter bumped), (4) modified canon/methods/persona-shaped-agent-runtime.md (new §8 Support Handoff-Insufficiency Signaling). Forward-looking URI reference to klappy://canon/methods/spawned-agent-session-substrate-options remains in frontmatter — that doc lands in PR A scope (substrate rename + substrate options); both PRs can land in either order, references resolve once both are merged. Mode-separated-conversations doc still needs Resolution-mode section (deferred follow-up). |
| 2 | +D Decision 5 5 strong Sessions Mirror Modes promoted to Tier-1 principle Each epistemic mode earns its own session because the structural blindness that makes a creator unable to validate their own work, a critic unable to remediate their own findings, and a planner unable to honestly execute their own plan is the same blindness in three different shapes. The principle generalizes critic-cannot-be-resolver (klappy://canon/constraints/critic-cannot-be-resolver) and verification-requires-fresh-context (klappy://canon/principles/verification-requires-fresh-context) to a universal claim: every gate between modes is a context boundary, and every context boundary deserves a fresh session. Five mode-bound roles named — explorer, planner, builder, validator, resolver — each mapping 1:1 to a canonical mode. |
| 3 | +D Decision 5 5 strong Mode Transitions Require Encoded Handoff promoted to Tier-1 constraint Every transition between modes requires two artifacts: a journal entry recording the transition (universal, no exception) and a transition-specific minimal handoff (explorer→planner: synthesis ledger; planner→builder: plan with assumptions and scope; builder→validator: artifact + claims; validator→resolver: findings with dispositions; resolver→re-validator: revised artifact + remediation summary). Reversion and skip remain permitted with explicit acknowledgment per existing canon. New: operator override added as third permitted deviation — explicit runtime declaration of mode collapse under declared urgency, with stated reason, acknowledged risks, and journal entries at session start and end. |
| 4 | +D Decision 5 5 strong Resolution promoted to fifth canonical epistemic mode The four-mode canon already implied a fifth phase: "fixes belong to iteration, which is a fresh execution pass scoped by validation findings." This patch makes the implied mode explicit. Resolution sits between Validation and a re-validation pass. Purpose: produce a revised artifact scoped strictly by validation findings. Truth condition: addresses findings without expanding scope. Primary risk: scope creep — using remediation as cover for redesign. Non-Collapse Rule extended with three resolution-specific invalid moves. Frontmatter date bumped to 2026-05-10; epoch to E0008.5; stability stays semi_stable since the addition is canonical implication-made-explicit, not a new claim about how truth is formed. |
| 5 | +D Decision 5 5 strong Handoff-Insufficiency Signaling added as runtime feature #8 The runtime cannot guarantee handoff quality — that is upstream encoding-norm discipline. But the runtime can ensure that bad handoffs are detectable. A receiving session that cannot proceed on an insufficient handoff returns a structured "handoff_insufficient" outcome with named missing items and a proposed_resolution (request fresh upstream session / request clarification / operator override). The runtime records the insufficiency in the journal, blocks the downstream handoff, and surfaces the resolution path to the consumer. Audit trails accumulate; chronic insufficiency at a specific gate signals norm gaps that the project's encoding discipline must address. |
| 6 | +L Learning 5 5 strong The architecture only pays back when handoffs preserve dynamic types Operator surface insight: "If encodings and other handoffs are lacking and of poor quality or missing key dynamic types any form of missing details and crucial context excluded will cause problems that may not outweigh the benefit of exclusive sessions." Handoff quality is the architecture's load-bearing assumption. Specific failure modes named: lossy encoding (decisions captured but not live tensions); missing dynamic types (static plan but not live state); missing crucial context that felt obvious to the encoder but wasn't written down because everyone-on-the-project-knew. When these dominate, session-per-mode becomes worse than mode-collapse: orchestration cost paid in full and receiving sessions operating on distorted views. Failure-mode sections added to sessions-mirror-modes principle (substantive), encoded-handoff constraint (pointer + summary), and persona-shaped-agent-runtime (handoff-insufficiency signaling feature). |
| 7 | +C Constraint 5 5 strong Encoding norms must emerge from observed failures The constraint codifies handoff PRESENCE; QUALITY is upstream encoding-norms discipline that the constraint requires the existence of but cannot measure. Type-system / runtime-validation pattern: encoded-handoff constraint catches some classes of failure statically (handoff exists with minimum content); handoff-insufficiency signaling catches the rest at runtime (receiving session structurally cannot do its job). A project that observes its planner-role sessions routinely flagging explorer-handoffs as insufficient on a specific dimension has evidence that its synthesis-ledger encoding norms need extension. The runtime surfaces failure patterns; the project refines norms. |
| 8 | +O Observation 5 5 strong Parallelism patterns named as four cases with one forbidden Within-mode parallelism (multiple explorers, multiple validators with different lenses, multiple builders on independent scope) — encouraged. Multi-participant single session (builder + assistant pair-programming) — allowed within a mode. Cross-mode on the same artifact — FORBIDDEN; the only pattern blocked. Cross-mode on different artifacts — fully independent, default mode of a productive multi-agent system. Distinguishing question: are these sessions operating on the same artifact in different modes simultaneously? If yes, forbidden; if no, allowed. Runtime enforces the forbidden pattern by refusing validator sessions on artifacts whose builder hasn't produced an encoded handoff. |
| 9 | +H Handoff 5 5 strong PR #187 update — five new sections + new principle + new constraint + epistemic-modes Resolution patch klappy.dev branch claude/runtime-canon-acd1 / PR #187 picks up: (1) new canon/principles/sessions-mirror-modes.md (Tier-1 principle, 25kb), (2) new canon/constraints/mode-transitions-require-encoded-handoff.md (Tier-1 constraint, 21kb), (3) modified canon/definitions/epistemic-modes.md (Resolution as fifth mode, Non-Collapse Rule extended, frontmatter bumped), (4) modified canon/methods/persona-shaped-agent-runtime.md (new §8 Support Handoff-Insufficiency Signaling). Forward-looking URI reference to klappy://canon/methods/spawned-agent-session-substrate-options remains in frontmatter — that doc lands in PR A scope (substrate rename + substrate options); both PRs can land in either order, references resolve once both are merged. Mode-separated-conversations doc still needs Resolution-mode section (deferred follow-up). |
0 commit comments