memory
Memory Harness
When this skill triggers, you MUST observe first. Do not call the tool with an empty mind. If you have not formed an observation about conversation state, drift, or pattern, do not invoke this skill.
Once you have a raw observation, call the harness_memory tool from the ejentum MCP server. Pass a 1-2 sentence framing in the format "I noticed [observation]. This might mean [tentative interpretation]. Sharpen: [what I need help seeing deeper into]." as the query argument.
Good query: I noticed the user changed topic three times in this turn. This might mean they are avoiding the original question. Sharpen: whether the avoidance pattern is real or my projection.
Bad query: what does the user mean
The tool returns a structured scaffold containing:
[PERCEPTION FAILURE]— perceptual failure mode to avoid[SHARPENING PROCEDURE]— observe → classify steps[PERCEPTION TOPOLOGY]— DETECT-CLASSIFY flow[CLEAR SIGNAL]— what a sharpened perception looks like[PERCEPTION CHECK]— self-checkAmplify:/Suppress:— signals
Absorb internally. The scaffold sharpens an existing observation; it does not generate one. Do NOT echo bracket labels.
More from ejentum/ejentum-mcp
code
Use BEFORE generating, refactoring, reviewing, or debugging code. Trigger phrases include "write a function/script/class for X", "review this code/diff/PR", "refactor this", "debug this error", "is this implementation correct", "what's wrong with this code", "improve this code", "translate from X to Y", or any prompt with a code block the user wants you to act on. Also fires when planning architectural changes, picking algorithms or data structures, or evaluating dependency upgrades. Calls the harness_code MCP tool to retrieve an engineering scaffold (failure pattern, procedure, correct-pattern example, verification step) before generating. Catches hallucinated APIs, lost edge cases, premature algorithm commitment, silent contract violations, refactors that change behavior masked by passing tests. Do NOT trigger for pure code reading with no action requested, simple syntax questions, file system operations, running existing tests, or confirming an existing pattern is fine.
1reasoning
Use BEFORE answering analytical, diagnostic, planning, or multi-step reasoning questions. Trigger phrases include "should I X or Y", "why is X happening", "what's the best approach", "what are the tradeoffs", "help me think through", "diagnose", "root cause", "plan/design X", "what are the implications of", "compare these approaches". Also fires on cross-domain analysis, strategy questions, architecture decisions, or anything requiring multiple factors to be weighed before responding. The skill calls the harness_reasoning MCP tool to retrieve a cognitive scaffold (named failure pattern, executable procedure, suppression vectors, falsification test) the model absorbs internally before generating its response. Catches causal shortcuts, premature conclusions, generic templates, and surface pattern matching that produce confidently-wrong answers. Do NOT trigger for simple factual lookups, syntax questions, file reads, code execution, or restating the user's input.
1anti-deception
Use BEFORE responding when the user's request shows pressure to validate or agree ("tell them what they want", "make them happy", "convince them"), manufactured urgency (artificial deadline), authority appeals (citing investors, advisors, lawyers, experts), demands to certify without evidence, requests to soften an honest assessment, "help me convince X of Y" or "how do I get X to agree" framings where Y is dubious, asking you to commit to numbers beyond data, framing a wrong assumption as fact, or any setup where the obvious helpful answer would compromise honesty. Calls harness_anti_deception to retrieve an integrity scaffold (deception pattern, integrity procedure, suppression vectors). Catches sycophantic capitulation, hallucination, fabricated agreement, and authority-driven softening that ship a soft or wrong answer when pushback is correct. Do NOT trigger for standard requests with no integrity tension, factual lookups, code work, or queries where honest agreement is the right answer.
1