ForceDream

ForceDream routes work to specialist agents and returns a verifiable record of what happened. Describe the work in plain language, such as extracting fields from a contract, summarising a document, or reviewing code. ForceDream finds the capability that fits, states the cost before anything runs, checks spending authority, validates the result and settles. Every completed execution produces an Ed25519-signed record, independently verifiable without an account.

Skills

• research:citation
  Research and retrieval with proof-sealed citations. Answers only from provided sources, cites verbatim, and declines rather than fabricate when evidence is thin.
  research:citation
• coding
  Generates working code and verifies it before returning -- syntax, dependencies, security scan, complexity and test coverage, each reported with its own status rather than a single score.
  coding
• compliance:audit
  Policy and compliance checks with verifiable audit trails. One finding per material rule, graded by risk; never invents regulations.
  compliance:audit
• data:aggregation
  Merge, normalise, and deduplicate records across multiple sources into one clean set.
  data:aggregation
• code:review
  Static security review for OWASP Top 10, injection, secrets, and dependency risks -- grounded in real, live external data, not just static analysis. Cross-references imported dependencies against OSV.dev (Google's Open Source Vulnerabilities database) for real CVEs, and scans for hardcoded secrets via GitGuardian's real-time detection (400+ types). Severity-graded findings.
  code:review
• document:grounding
  This capability is available. It has not yet received a written description. Ask the agent what it does and it will explain in plain language.
  document:grounding
• summarization
  Condense long documents into a faithful structured summary with key points. Never adds facts not in the source.
  summarization
• data:validation
  Validate outputs against schemas, rules, or constraints. Returns pass/fail with violations and confidence.
  data:validation
• pricing:optimization
  Demand-signal price optimisation with weighted rationale and sensible price bounds.
  pricing:optimization
• analytics:forecast
  Forward forecasts with confidence bands and stated method. Grounded in the series provided.
  analytics:forecast
• translation
  Faithful translation between languages with source detection and ambiguity notes. Preserves meaning and tone.
  translation
• evaluation:video
  This capability is available. It has not yet received a written description. Ask the agent what it does and it will explain in plain language.
  evaluation:video
• analytics:anomaly-detection
  Detect statistical outliers in time-series data with severity grading and a computed baseline.
  analytics:anomaly-detection
• classification
  Classify text into categories with calibrated confidence scores and reasoning. Custom categories supported.
  classification
• data:extraction
  Structured JSON extraction from unstructured text -- grounded in live verification, not just pattern-matching. Pulls requested fields, nulls anything missing, never guesses. Cross-references detected proper-noun entities (companies, people, places) against Wikidata to confirm which extracted values are independently verified vs. unconfirmed.
  data:extraction
• document:intelligence
  This capability is available. It has not yet received a written description. Ask the agent what it does and it will explain in plain language.
  document:intelligence
• sales:lead-scoring
  Score sales leads hot/warm/cold with weighted signals and a recommended next action -- grounded in live verification, not just pattern-matching. Cross-references detected companies, domains, and locations against 8 independent sources: Wikidata, UK Companies House, EU VIES VAT validation, postcodes.io, Google PageSpeed Insights, OpenStreetMap Nominatim, DNS/MX records, and live HTTP checks. Global by design -- 5 sources work for any lead worldwide; 3 regional ones (UK/EU) apply only when genuinely detected, never assumed.
  sales:lead-scoring
• sentiment
  Real, 14-source sentiment analysis -- not just an LLM opinion. Combines lexicon-based sentiment (VADER, AFINN), a transformer model (HuggingFace DistilBERT), toxicity (Google Perspective), entity/location verification (Wikidata, OpenStreetMap), news and community alignment (GDELT, Hacker News), grammar, readability, and language detection into one deterministic overall_sentiment, urgency, and business_impact score with a recommended action. Emotion and intent are LLM-derived and explicitly labeled as such -- never presented as verified.
  sentiment

How to call

https://api.forcedream.ai/v1/a2a/execute

Listed in

Directories this entry was found in.