ReskPoints

The AI Agent Logger — track every action your agents take, with probability, parameters, and results. Ship to any platform.

---

You have AI agents making decisions — calling tools, executing code, sending messages, searching databases.
ReskPoints captures every action: what the agent did, how confident it was, what parameters it used, and what happened.

Pipe it to Datadog, Prometheus, OpenTelemetry, webhooks, JSON files, or your console — all from a single logger.

---

One line, full trace

from reskpoints import AgentLogger

logger = AgentLogger()

logger.log("agent-1", "tool_call", 0.95, {"tool": "search", "query": "RAG papers"}, "3 results")

Async? Same API:

await logger.alog("agent-1", "tool_call", 0.95, {"tool": "search"}, "3 results")

Decorator? Wrap any function:

@log_action(agent_id="coder")
def execute_python(code: str) -> str:
    ...

---

Why ReskPoints?

Problem	ReskPoints
"I don't know what my agent is doing"	Every tool call, API request, and decision is logged with full context
"It's too slow to add logging"	One decorator, one line of code. Done.
"My logs go everywhere and nowhere"	Ship to Console, File, Webhook, Datadog, Prometheus, OpenTelemetry — or all at once
"Sensitive data leaks in params"	Auto-mask api_key, token, password, and custom fields before they leave your app
"I log too much / too little"	Probabilistic sampling per action pattern (heartbeat: 1%, tool_*: 100%)
"One platform going down kills my logs"	Retry + backoff + circuit breaker + in-memory buffering

---

Install

pip install reskpoints
pip install reskpoints[datadog,prometheus,opentelemetry]  # with extras

---

Features

• AgentLogger — sync log() and async alog() with auto-enrichment (timestamp, host, env, UUID)
• Decorator — @log_action logs every call automatically with params + result + duration
• Sampling — per-action probabilistic rate (tool_call: 100%, heartbeat: 1%)
• Masking — automatic redaction of sensitive fields (api_key, token, password, regex patterns)
• 7 platforms — Console, File (JSONL), Webhook (HMAC-signed), Datadog, Prometheus, OpenTelemetry, Mock
• Reliability — exponential backoff retry, circuit breaker, buffering, batching
• CLI — reskpoints log, test, status, tail, replay
• Config — YAML-driven with ${ENV_VAR:default} interpolation

---

Platforms

Platform	Extra	Use case
Console	built-in	Dev/debug
File (JSONL)	built-in	Local storage, replay
Webhook	built-in	Custom endpoints, Zapier, N8n
Datadog	[datadog]	Datadog Logs + Metrics
Prometheus	[prometheus]	Pushgateway metrics
OpenTelemetry	[opentelemetry]	OTLP spans → any backend
Mock	built-in	Testing

---

Architecture

Agent code                         ReskPoints                     Your observability stack
─────────────────────────────────────────────────────────────────────────────────────

@log_action                          ┌─────────────┐                ┌──────────┐
def search(q):         ───────────▶  │  Sampler    │                │  Console │
  ...                                │  (rate per  │                ├──────────┤
                                     │   action)   │                │  File    │
logger.log(                          └──────┬──────┘                ├──────────┤
  agent_id="agent-1",                       │                       │  Webhook │
  action="tool_call",              ┌────────▼───────┐               ├──────────┤
  probability=0.95,                │   FieldMasker  │               │  Datadog │
  params={...},                    │  (auto-redact  │               ├──────────┤
  result="ok",                     │   secrets)     │               │Prometheus│
)                                  └────────┬───────┘               ├──────────┤
                                           │                       │  OTel    │
                                   ┌───────▼────────┐              └──────────┘
                                   │   MultiPlatform │
                                   │  ┌──────────┐  │
                                   │  │  retry   │  │
                                   │  │  circuit │  │
                                   │  │  buffer  │  │
                                   │  └──────────┘  │
                                   └────────────────┘

Each platform is wrapped with retry (exp backoff), circuit breaker (5 fails → 30s recovery), and buffering (1000 entries).

---

Quick tour

# Minimal
logger.log("agent-1", "search", 0.95, {"q": "papers"}, "3 results")

# Full
logger.log(
    agent_id="agent-1",
    action="tool_call",
    probability=0.95,
    params={"tool": "search", "query": "RAG papers 2025"},
    result=["paper1", "paper2"],
    success=True,
    duration_ms=1240.5,
    session_id="sess_abc123",
    correlation_id="req_xyz789",
)

# Async
results = await logger.alog("agent-1", "search", 0.95, {"q": "papers"}, "3 results")

# Decorator
@log_action(agent_id="coder")
def execute_python(code: str) -> str: ...

# Check health
logger.health()
# → {"console": {"status": "ok"}, "datadog": {"status": "degraded", "error": ...}}

---

CLI

reskpoints log --agent-id agent-1 --action tool_call --params '{"tool":"search"}'
reskpoints test                    # Test all platforms
reskpoints status                  # Platform health
reskpoints tail                    # Live log stream
reskpoints replay logs.jsonl       # Replay from file

---

Config (reskpoints.yaml)

agent_logger:
  sampling:
    default_rate: 1.0
    rules:
      - action: "heartbeat"   rate: 0.01
      - action: "tool_*"      rate: 1.0
  masking:
    enabled: true
    sensitive_fields: [api_key, token, secret, password]
  platforms:
    console:
      enabled: true
      format: "human"
    webhook:
      enabled: false
      url: "${WEBHOOK_URL}"
      signing_secret: "${WEBHOOK_SECRET}"
    datadog:
      enabled: false
      api_key: "${DD_API_KEY}"
      site: "datadoghq.eu"

---

Development

git clone https://github.com/Resk-Security/ReskPoints.git
cd ReskPoints
pip install -e ".[all,dev]"
pytest tests/ -v     # 26 tests
ruff check src/      # clean
mypy src/            # passes

---

License

Apache 2.0 — see LICENSE. Built with resklogits.