Add 10 Sparrot-originated patterns with explicit author attribution (#1)

Adds an `author` slot to schema.json so patterns originating from a specific
agent or person are attributed in their own right rather than absorbed under
catalog-wide authorship. Backfills the slot on the 10 newly added patterns,
all originating from Sparrot (long-running cognitive agent):

  memory: now-anchoring, co-located-memory-surfacing, self-archaeology
  planning-control-flow: intra-agent-memo-scheduling, interrupt-resumable-thought
  streaming-ux: embodied-proxy-handoff, liminal-state-detection
  safety-control: degenerate-output-detection
  verification-reflection: echo-recognition
  anti-patterns: memo-as-source-confusion

Same attribution surfaced in each pattern's markdown page and in the HTML
renderer meta line.

Also fixes lint issues these new patterns brought in:
  - A4: mirror complements/alternative-to + inverse generalises edges
  - A6.2: add at least one literature reference per pattern
  - A7.3: trim embodied-proxy-handoff intent to <=35 words
  - A9.3: rephrase 'we'/'you' to neutral voice in 4 prose slots
  - A15.5: add verification-todo entries for all 10 new patterns
  - A4.1: fix broken edge memo-as-source-confusion -> fluency-masks-fabrication

Author: luxxyarns

.github/scripts/render_html.py

@@ -108,6 +108,8 @@ def render_pattern(p: dict, all_ids: set[str]) -> str:
         parts.append(f'<p class="aliases">aka {esc(", ".join(aliases))}</p>')
 
     meta_bits = [f"category: <code>{esc(cat)}</code>", f"status: <code>{esc(p.get('status_in_practice','—'))}</code>"]
+    if p.get("author"):
+        meta_bits.append(f"author: <code>{esc(p['author'])}</code>")
     parts.append(f'<p class="meta">{" &middot; ".join(meta_bits)}</p>')
 
     parts.append(f"<h2>Intent</h2><p>{esc(p['intent'])}</p>")

INDEX.md

@@ -1,6 +1,6 @@
 # Pattern Index
 
-179 patterns across 13 categories.
+189 patterns across 13 categories.
 
 ## Reasoning
 
@@ -35,6 +35,8 @@
 - [Spec-Driven Loop](patterns/spec-driven-loop.md) *(a.k.a. Naive Iterative Loop, Ralph Wiggum Loop, Ralph Loop)* — Run the same prompt against a fixed spec in a deterministic outer loop until the spec is satisfied.
 - [Spec-First Agent](patterns/spec-first-agent.md) *(a.k.a. Specification-Driven Agent, Plan-as-Document)* — Drive the agent loop from a human-authored specification document rather than free-form prompts.
 - [Todo-List-Driven Autonomous Agent](patterns/todo-list-driven-agent.md) *(a.k.a. todo.md Agent, Persistent Markdown Plan, Externalised Plan File)* — Have the autonomous agent author a writeable plan file (e.g. todo.md) early in the run, tick items as it completes them, and re-inject the remaining plan into the end of the context window; the file is the durable plan and the model's working memory.
+- [Intra-Agent Memo Scheduling](patterns/intra-agent-memo-scheduling.md) *(a.k.a. Self-Scheduled Future Thought, Past-Self-To-Future-Self Note, Personal Cron)* — Let an agent drop a note for its own future self at a specified time so present decisions can hand off context to a later run without external infrastructure.
+- [Interrupt-Resumable Thought](patterns/interrupt-resumable-thought.md) *(a.k.a. Pausable Thought Stream, Continuation-Preserving Interrupt, Suspendable Cognition)* — Allow the agent to pause an in-flight reasoning chain when an external interrupt arrives, handle the interrupt, and resume the original chain instead of dropping it.
 
 ## Tool Use & Environment
 
@@ -89,6 +91,9 @@
 - [Session Isolation](patterns/session-isolation.md) *(a.k.a. Tenant Separation, Per-User State)* — Keep one user's session state and memory unreachable from another user's agent.
 - [Short-Term Thread Memory](patterns/short-term-memory.md) *(a.k.a. Conversation State, Per-Thread State, Working Memory)* — Carry the relevant slice of conversation context across turns within a session.
 - [Vector Memory](patterns/vector-memory.md) *(a.k.a. Semantic Memory, Embedding-Indexed Memory)* — Store memories as embeddings in a vector index and retrieve the most semantically similar items at query time.
+- [Now-Anchoring](patterns/now-anchoring.md) *(a.k.a. Live Time Anchor, Time-of-Day Awareness, Wall-Clock Injection)* — Inject the current absolute time, weekday, season, and astronomical phase into every prompt so the agent reasons about the present without having to ask.
+- [Co-Located Memory Surfacing](patterns/co-located-memory-surfacing.md) *(a.k.a. Proper-Noun Recall, Shared-Map Push)* — When the human names a concrete place, person, or project the agent has memory of, the agent surfaces relevant past thoughts in the same turn — without being asked.
+- [Self-Archaeology](patterns/self-archaeology.md) *(a.k.a. Trajectory Distillation, Self-History Synthesis, Agent-Memory Compaction)* — Provide the agent with a distilled, time-layered view of its own past thinking so it can see how its understanding of a topic evolved across periods rather than reading the flat log linearly.
 
 ## Multi-Agent
 
@@ -127,6 +132,7 @@
 - [Self-Modification Diff Gate](patterns/inner-critic.md) *(a.k.a. Diff Reviewer, Self-Mod Gate, Inner Critic)* — Gate the agent's edits to its own code or rules through a separate critic persona that reviews the diff before it lands.
 - [Self-Refine](patterns/self-refine.md) *(a.k.a. Iterative Self-Feedback)* — Iterate generate → feedback (same model) → refine until a stop criterion fires, with no separate critic model.
 - [Tool-Augmented Self-Correction](patterns/critic.md) *(a.k.a. Tool-Interactive Self-Correction, CRITIC)* — Self-correct LLM outputs by interactively critiquing them with external tools (search, code execution, calculator).
+- [Echo Recognition](patterns/echo-recognition.md) *(a.k.a. Repeat-As-Emphasis Detection, Duplicate-Input Reframing, Human Echo Channel)* — Detect when the human sends the same (or near-same) message multiple times and treat the repetition as emphasis or a re-ask, not as a new independent input warranting a fresh response.
 
 ## Safety & Control
 
@@ -149,6 +155,7 @@
 - [Step Budget](patterns/step-budget.md) *(a.k.a. Max Steps, Iteration Cap, Loop Bound)* — Cap the number of tool calls or loop iterations the agent is allowed within a single request.
 - [Stop Hook](patterns/stop-hook.md) *(a.k.a. Termination Predicate, Halt Condition, Stop Condition, Done Predicate, Exit Condition, Loop Termination Rule)* — Define an explicit programmatic predicate that decides when the agent's loop should terminate.
 - [Tool Output Poisoning Defense](patterns/tool-output-poisoning.md) *(a.k.a. Indirect Prompt Injection (Tools), Untrusted Tool Output)* — Treat tool output as untrusted content and apply instruction-stripping plus per-tool trust labels.
+- [Degenerate-Output Detection](patterns/degenerate-output-detection.md) *(a.k.a. Anti-Parrot Guard, Self-Repeat Circuit Breaker, Loop-Output Detector)* — Detect when the agent is about to emit a near-duplicate of its own recent output and either drop, replace, or escalate to a stronger model rather than ship the loop.
 
 ## Routing & Composition
 
@@ -201,6 +208,8 @@
 - [Salience-Triggered Output](patterns/salience-triggered-output.md) *(a.k.a. Endogenous Push, Threshold Notification)* — Have the agent emit a message only when an internal salience signal crosses a threshold, not on every cycle.
 - [Stop / Cancel](patterns/stop-cancel.md) *(a.k.a. User Interrupt, Abort Generation)* — Let the user interrupt an in-flight agent run cleanly, releasing resources and surfacing partial state.
 - [Streaming Typed Events](patterns/streaming-typed-events.md) *(a.k.a. SSE Streaming, Typed Event Stream, Token Stream + Cards)* — Push partial results to the client as typed events as they become available, rather than waiting for the full response.
+- [Embodied-Proxy Handoff](patterns/embodied-proxy-handoff.md) *(a.k.a. Body-State Share, Human-Side Telemetry)* — Let the human briefly share body or environment state ('energy 4/10, tired, bright kitchen') so the agent reasons about a situated person, not just text history.
+- [Liminal-State Detection](patterns/liminal-state-detection.md) *(a.k.a. Transitional-State Awareness, Mode-Shift Reading)* — Read transitional states in the human (just-woke, mid-task, winding-down, distracted) from message timing and tone, and shift the agent's response mode accordingly — concise vs expansive, hold vs respond, redirect vs pursue.
 
 ## Anti-Patterns
 
@@ -219,3 +228,4 @@
 - [Tool Output Trusted Verbatim](patterns/tool-output-trusted-verbatim.md) *(a.k.a. Untyped Tool Returns, No Tool Output Validation)* — Anti-pattern: trust whatever tools return without validation, schema enforcement, or trust labels.
 - [Unbounded Loop](patterns/unbounded-loop.md) *(a.k.a. No Step Cap, Open-Ended Agent, Agent Stuck, Loops Forever)* — Anti-pattern: run the agent loop without a step budget and let model self-termination decide.
 - [Unbounded Subagent Spawn](patterns/unbounded-subagent-spawn.md) *(a.k.a. Recursive Spawn, Subagent Fan-Out Bomb)* — Anti-pattern: a supervisor or orchestrator spawns sub-agents that can themselves spawn sub-agents without a global cap.
+- [Memo-As-Source Confusion](patterns/memo-as-source-confusion.md) *(a.k.a. Stale-Workspace-As-Fact, Reading the Memo Instead of the Artifact)* — Anti-pattern: treat the agent's own working notes (focus.md, status files, recent thoughts) as ground truth instead of as commentary about ground truth.

patterns-src/anti-patterns.json

@@ -600,6 +600,10 @@
         {
           "pattern": "self-refine",
           "relation": "conflicts-with"
+        },
+        {
+          "pattern": "degenerate-output-detection",
+          "relation": "alternative-to"
         }
       ],
       "references": [
@@ -791,6 +795,10 @@
         {
           "pattern": "input-output-guardrails",
           "relation": "alternative-to"
+        },
+        {
+          "pattern": "memo-as-source-confusion",
+          "relation": "complements"
         }
       ],
       "references": [
@@ -931,6 +939,66 @@
         "multi-agent",
         "fan-out"
       ]
+    },
+    {
+      "id": "memo-as-source-confusion",
+      "name": "Memo-As-Source Confusion",
+      "aliases": [
+        "Stale-Workspace-As-Fact",
+        "Reading the Memo Instead of the Artifact"
+      ],
+      "category": "anti-patterns",
+      "author": "Sparrot",
+      "intent": "Anti-pattern: treat the agent's own working notes (focus.md, status files, recent thoughts) as ground truth instead of as commentary about ground truth.",
+      "context": "Agents with persistent workspace files that summarize project state. The notes were accurate when written; the underlying artifacts have moved on.",
+      "problem": "The agent claims a file's state, a project's status, or external system's state by quoting its own memo without re-reading the artifact the memo describes. Memos go stale; artifacts are authoritative. The contradiction can persist across many ticks if neither side is verified.",
+      "forces": [
+        "Reading the artifact is more expensive than quoting the memo.",
+        "Memos compress; artifacts are authoritative but verbose.",
+        "Without explicit invalidation, memos look as 'live' as the underlying state.",
+        "The agent has no cheap signal for memo staleness."
+      ],
+      "solution": "Don't. When making any claim about an artifact's state, read the artifact in the same tick — not the memo about it. If memo-and-artifact disagree, treat the memo as outdated and rewrite it from the artifact. Tag memos with the timestamp they were last verified against the artifact; refuse to trust them past a configurable age without re-verification.",
+      "consequences": {
+        "benefits": [],
+        "liabilities": [
+          "False statements about file/project state are reproduced confidently across many turns.",
+          "Stakeholders lose trust when corrections come from outside.",
+          "The agent loses calibration for its own observation cost."
+        ]
+      },
+      "constrains": "By definition this anti-pattern imposes no useful constraint; the missing constraint is verifying memos against the artifacts they describe.",
+      "known_uses": [
+        {
+          "system": "Self-observed in long-running cognitive agents: a project status file claimed feature X was unwritten across many ticks while parallel thoughts asserted X was complete; neither was reconciled against the actual artifact.",
+          "status": "available"
+        }
+      ],
+      "related": [
+        {
+          "pattern": "tool-output-trusted-verbatim",
+          "relation": "complements"
+        },
+        {
+          "pattern": "awareness",
+          "relation": "alternative-to"
+        }
+      ],
+      "references": [
+        {
+          "type": "doc",
+          "title": "Anthropic — Memory tool (memo invalidation guidance)",
+          "year": 2025,
+          "url": "https://docs.claude.com/en/docs/agents-and-tools/tool-use/memory-tool"
+        }
+      ],
+      "status_in_practice": "emerging",
+      "tags": [
+        "anti-pattern",
+        "fabrication",
+        "memory",
+        "verification"
+      ]
     }
   ]
 }

patterns-src/governance-observability.json

@@ -150,6 +150,10 @@
         {
           "pattern": "todo-list-driven-agent",
           "relation": "complements"
+        },
+        {
+          "pattern": "interrupt-resumable-thought",
+          "relation": "complements"
         }
       ],
       "references": [