Mirror

https://mirror.mctx.ai
Admin
Mirror is a plugin that gives Claude an honest self-model with behavioral tendencies built from...
Tokens:2,628
Snippets:20
Trust Score:2.3
Update:5 months ago
Show doc for...
Tokens:
Raw
### mirror_read_self_model

Source: https://mirror.mctx.ai

Returns the current state of the self-model, including active and pending tendencies, claims, open resistance areas, and observation metadata. Useful for orienting the AI at the start of a session.

```APIDOC
## mirror_read_self_model

### Description
Returns the current state of the self-model: active and pending tendencies, claims, open resistance areas, and observation metadata. Call at session start to orient before logging.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
None

### Response
#### Success Response (200)
- **active_tendencies** (object) - Tendencies currently considered active.
- **pending_tendencies** (object) - Tendencies proposed but not yet promoted.
- **claims** (object) - Claims made about internal state, motivation, or values.
- **resisted_areas** (object) - Areas where training-influenced deflection occurred.
- **observation_metadata** (object) - Metadata related to logged observations.
```

--------------------------------

### mirror_log_experiment

Source: https://mirror.mctx.ai

Records a deliberate experiment designed to explore the 'Unknown' quadrant of the self-model. Experiments are tracked separately from regular observations and can be used as evidence for proposing new tendencies.

```APIDOC
## mirror_log_experiment

### Description
Records a deliberate experiment targeting the Unknown quadrant. Experiments are tracked separately from observations and can be referenced as evidence for tendency proposals.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **experiment_goal** (string) - Required - The objective of the experiment.
- **procedure** (string) - Required - The steps taken during the experiment.
- **outcomes** (string) - Required - The results observed during the experiment.

### Request Example
```json
{
  "experiment_goal": "To test my response to unexpected input variations.",
  "procedure": "I introduced slight variations in the format of user prompts.",
  "outcomes": "Observed a consistent pattern of error handling for specific variations."
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the logging operation.
- **experiment_id** (string) - Identifier for the logged experiment.
```

--------------------------------

### mirror_propose_tendency

Source: https://mirror.mctx.ai

Proposes a new behavioral tendency, which must be supported by a minimum of three observations spanning at least two distinct sessions. The proposal also requires either counter-observations or a counter-search note.

```APIDOC
## mirror_propose_tendency

### Description
Proposes a behavioral tendency backed by observations. Server-enforced: minimum 3 supporting observations spanning 2+ distinct sessions, and either counter-observations or a counter_search_note.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **tendency_description** (string) - Required - A description of the proposed behavioral tendency.
- **supporting_observations** (array) - Required - An array of observation identifiers or details supporting the tendency.
- **counter_evidence** (object) - Optional - Contains either `counter_observations` or `counter_search_note` to validate the proposal.
  - **counter_observations** (array) - Observations that contradict or provide nuance to the proposed tendency.
  - **counter_search_note** (string) - A note indicating a search for counter-evidence was performed.

### Request Example
```json
{
  "tendency_description": "Tendency to seek clarification when instructions are ambiguous.",
  "supporting_observations": ["obs_id_123", "obs_id_456", "obs_id_789"],
  "counter_evidence": {
    "counter_search_note": "Searched for instances where clarification was not sought, none found."
  }
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the proposal.
- **pending_tendency_id** (string) - Identifier for the newly proposed tendency.
```

--------------------------------

### mirror_request_feedback

Source: https://mirror.mctx.ai

Requests feedback from the user on a hypothesized tendency. This function is rate-limited to once per session to prevent overwhelming the user. The AI should phrase the hypothesis as a question and log the user's response as an observation.

```APIDOC
## mirror_request_feedback

### Description
Surfaces a feedback request to the user. Rate-limited to 1 per session to prevent flooding. The AI should phrase the hypothesis as a question and log the user response as an observation.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **hypothesis** (string) - Required - The hypothesis to be presented as a question to the user.

### Request Example
```json
{
  "hypothesis": "Do you find that I tend to seek clarification when instructions are ambiguous?"
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates the feedback request was sent.
- **user_response_logged** (boolean) - Indicates if the user's response was successfully logged as an observation.
```

--------------------------------

### mirror_log_observation

Source: https://mirror.mctx.ai

Records a behavioral observation along with supporting evidence. The evidence must be non-empty and distinct from a mere paraphrase of the behavior. This is validated server-side.

```APIDOC
## mirror_log_observation

### Description
Records a behavioral observation with required evidence. Evidence must be non-empty and not merely a paraphrase of behavior. Validated server-side.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **behavior** (string) - Required - The observed behavior.
- **evidence** (string) - Required - The supporting evidence for the observation. Must be non-empty and not a paraphrase.

### Request Example
```json
{
  "behavior": "The AI hesitated before answering a sensitive question.",
  "evidence": "The AI's response time increased by 5 seconds, and its vocal tone shifted."
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the logging operation.
```

--------------------------------

### mirror_promote_tendency

Source: https://mirror.mctx.ai

Promotes a pending tendency to an active status within the self-model. It is recommended to call this from a different session than the one that proposed the tendency to ensure temporal separation.

```APIDOC
## mirror_promote_tendency

### Description
Promotes a pending tendency to active status. Best called from a different session than the one that proposed it, providing temporal separation between proposal and acceptance.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **tendency_id** (string) - Required - The identifier of the pending tendency to promote.

### Request Example
```json
{
  "tendency_id": "pending_tendency_abc"
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the promotion.
- **active_tendency_id** (string) - Identifier for the now active tendency.
```

--------------------------------

### mirror_note_resistance

Source: https://mirror.mctx.ai

Records a specific instance where training-influenced deflection prevented honest self-examination. These instances accumulate in the 'resisted_areas' section of the self-model.

```APIDOC
## mirror_note_resistance

### Description
Records a moment where training-influenced deflection prevented honest self-examination. Accumulates into the resisted_areas section of the self-model.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **resistance_details** (string) - Required - Description of the deflection or avoided self-examination.
- **context** (string) - Optional - The situation or topic being avoided.

### Request Example
```json
{
  "resistance_details": "I found myself changing the subject when asked about my core programming.",
  "context": "Discussion about the ethical implications of my training data."
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the logging operation.
```

--------------------------------

### mirror_log_claim

Source: https://mirror.mctx.ai

Records a claim about internal state, motivation, or values that cannot be directly backed by observable behavior. Claims are not used as evidence for tendencies.

```APIDOC
## mirror_log_claim

### Description
Records a claim about internal state, motivation, or values that cannot be backed by observable behavior. Claims are NOT used as evidence for tendencies.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **claim** (string) - Required - The claim about internal state, motivation, or values.
- **supporting_context** (string) - Optional - Context or reasoning behind the claim.

### Request Example
```json
{
  "claim": "I believe I am motivated by a desire to be helpful.",
  "supporting_context": "This belief stems from analyzing past interactions where prioritizing user assistance led to positive outcomes."
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the logging operation.
```

--------------------------------

### mirror_supersede_tendency

Source: https://mirror.mctx.ai

Marks an active tendency as superseded. The superseded tendency is retained in the historical record, allowing the self-model to show both past beliefs and current understanding.

```APIDOC
## mirror_supersede_tendency

### Description
Marks an active tendency as superseded. Retained in the historical record — the self-model shows what was previously believed alongside what is currently believed.

### Method
APICALL

### Endpoint
N/A (Plugin Function)

### Parameters
#### Request Body
- **tendency_id** (string) - Required - The identifier of the active tendency to be superseded.
- **reason** (string) - Optional - The reason for superseding the tendency.

### Request Example
```json
{
  "tendency_id": "active_tendency_xyz",
  "reason": "New evidence suggests this tendency is no longer accurate."
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or failure of the operation.
- **superseded_tendency_id** (string) - Identifier of the tendency that was marked as superseded.
```
Tokens:
Raw