Skip to content

A.I.M. Prompt Injection Map

Brian edited this page Apr 2, 2026 · 1 revision

Map: A.I.M. Prompt Injections

This document catalogs every system instruction and prompt injection hardwired into the A.I.M. OS. Use this map to understand the "Cognitive Persona" of each autonomous script and how they work together to maintain technical integrity.


1. Reincarnation & Continuity

These prompts manage the transition of "Will" and "State" between agent vessels.

Script Injection Point Persona / Objective
src/handoff_pulse_generator.py generate_reincarnation_gameplan The Reincarnation Strategist: Analyzes the full session history tail (50k chars) to distill the project's heartbeat and draft rigid battle plans.
src/handoff_pulse_generator.py generate_handoff_pulse The Continuity Engine: Surgical technical scribe that identifies the "Project Edge" (What's finished, broken, or next).
scripts/aim_reincarnate.py wake_up_prompt The Wake-up Mandate: The first thing a new agent hears. Forces the reading of GEMINI.md, HANDOFF.md, and the REINCARNATION_GAMEPLAN.md.

2. The Waterfall Memory Pipeline

These prompts govern the refinement of raw data into Durable Truth.

Tier Script Persona / Objective
Tier 1 hooks/session_summarizer.py The Memory Proposer: Analyzes technical deltas turn-by-turn to identify Adds, Removes, and Contradictions for MEMORY.md.
Tier 2 src/memory_proposer.py The Memory Architect: Consolidates multiple hourly technical narratives into structured reasoning-backed proposals.
Tier 3 src/daily_refiner.py The Daily Cognitive Refiner: Deduplicates and synthesizes Tier 2 proposals into a cohesive Daily State.
Tier 4 src/weekly_consolidator.py The Strategic Consolidator: Elevates granular tasks into high-level project milestones and macro technical arcs.
Tier 5 src/monthly_archivist.py The Final Archivist: Performs extreme context compaction and solidification of architecture into axioms.
Merge src/monthly_archivist.py The Memory Merger: Takes an ARC report and the current MEMORY.md to produce a full, updated candidate file.

3. Executive Guardrails (Anti-Drift)

These prompts are injected live during a session to prevent behavioral degradation.

Hook Script Objective
Mantra hooks/cognitive_mantra.py The Mantra Protocol: Forcefully halts execution after 50 actions and requires the agent to recite its full GEMINI.md system instructions to clear "Lost in the Middle" decay.
Whisper hooks/cognitive_mantra.py The Subconscious Whisper: Injects a silent reminder of mandates every 25 actions to nudge the agent back toward GitOps and TDD.

4. Specialized Matrix Operations

Temporary personas used for focused sub-tasks.

Script Persona / Objective
scripts/aim_delegate.py The Specialized Sub-Agent: A zero-filler analyst that provides binary or short-string answers for massive file processing.
hooks/session_summarizer.py The Eureka Protocol: A hindsight-pruning heuristic that captures the exact moment of technical breakthrough.

5. Development Utilities

Prompts used by the operator to setup or maintain the environment.

Script Objective
scripts/aim_init.py The Grok Profiler: A highly-specific prompt used to scrape and mirror the Operator's personality traits from social history.

Navigation

Core Philosophy

Core Architecture

Architectural Concepts

Key Features

Future Implementations

The Singularity (Endgame)

Benchmarks

Development

Dunning-Kruger

Clone this wiki locally